fbpx

Top 5 Startups Leading in Drug Discovery using AI

If drug discovery is your bread and butter, you might already know that drug development often takes a decade of research and billions in investment before the drug reaches the market. In other words, the drug development process is complicated. Difficult. And a rather costly business. But all hope’s not lost yet.

To detangle the process of drug discovery. Artificial Intelligence and machine learning have come as a ray of hope for the pharmaceutical industry. AI solutions allow researchers to quickly design novel drugs that display the desired properties. Many companies are working on designing novel drug molecules using these advanced technologies.

 

Talking about novel drugs, every year, patents on these novel drugs get expired. As monitoring these drug patents are important to many pharma companies, we listed over 200 drug patents that are expiring between 2021 to 2025. If you want the information, fill out the form below:

Which are these companies and how are they using the power of AI and ML?

To find out, we dived into this technology and checked the companies providing unique solutions. We found some startups utilizing innovative ways and filing patents to solve the problems faced during the drug-discovery process. Many companies are collaborating with and acquiring different start-ups to become future leaders in this domain. In this article, we will be focusing on the start-ups that are leading and shaping this industry by employing the power of AI to discover the cure for serious human diseases.

How did we shortlist these start-ups?

The company has recently announced its first AI-designed molecule for treating cancer using the body’s own immune system, co-developed with Evotec. It will be entering human clinical trials soon. This happened following its 2020 announcement of bringing an AI-designed molecule for treating obsessive-compulsive disorder (OCD) to a phase 1 clinical trial in partnership with Japanese collaborator Sumitomo Dainippon Pharma.

Acquisition/Mergers and Collaborations

Exscientia has acquired Allcyte, an Austrian company developing an AI platform to predict how well cancer treatments will work in individual patients. Allcyte’s technology relies on deep learning to analyze the effects of various drugs on live samples of an individual patient’s tissue, rather than on artificial or animal models. By using allcyte’s platforms, Exscientia will be able to take a precision medicine approach to design drug molecules, ensuring that they’re even more effective at targeting tumor tissues than those designed with Exscientia’s technology alone.

Recently, the company also entered into a joint venture with GT Apeiron Therapeutics (Apeiron), a Shanghai-based company, to provide its AI-driven drug identification and design capabilities to accelerate the discovery of a therapeutic drug.

EQRx and Exscientia also team up on a discovery-through-commercialization to bring cheaper medicines to patients faster. As part of the deal, Exscientia will pick up discovery duties while EQRx will handle development and commercialization.

Where is the startup receiving its funding from?

In March 2021, Bill Gates-backed Oxford biotech raised $100M funding to discover new drugs using AI (source).

In April 2021, Softbank led a $525m round in British drug developer Exscientia (source).

Standigm

 

South Korean start-up, Standigm, founded in 2015, discovers drugs using AI, saving time and cost when compared to traditional methods. Their AI platform, Standigm BEST, is intended to explore latent chemical space to generate novel compounds with desired properties. Other than chemical data, they also analyze biomedical literature to speed up de novo drug design.

Once candidates have been identified, Standigm Insights provides biological interpretations to discover pathways, and therapeutics patterns and prioritize potential targets. The startup’s solutions eliminate uncertainty in the drug discovery process to save time and costs during development.

 

 Image credit – Standigm

In a nutshell, Standigm has achieved the automation of the whole drug discovery process based on their AI platforms, including Standigm ASK™ for target discovery, Standigm BEST™ for lead design, and Standigm Insight™ for drug repurposing. To date, Standigm has run 22 in-house or collaborative drug pipelines using ‌workflow AI technology.

The startup has filed a total of 10 patents. Given below are some of the interesting patents related to their innovative technology –

  1. KR2018022537A – This patent is mainly focusing on a method that utilizes machine learning in predicting the effects of the drug combination. This method uses a computer algorithm to collect various data –

  • cell-related data
  • drug-related data and,
  • drug/cell correlation-related data

All this collected data enables the prediction of the therapeutic efficacy of a new combination drug when synergic effects are expected according to the drug combinations. This method helps in finding out which targets or proteins are affected by which drugs. And how they interact if used in combination for the treatment purpose.

  1. KR2020145835A – This patent talks about a thioridazine composition used to treat or prevent metabolic liver disease, steatosis, inflammation, reduces fibrosis, non-alcoholic fatty liver disease, non-alcoholic steatohepatitis, etc. Further, the thioridazine molecule is selected through a screening method using artificial intelligence (AI) deep learning technology after confirming its therapeutic efficacy.
  2. WO2021080295A1 – This patent describes a method and apparatus for generating an effective compound structure by using the learning data of existing biologically active structures. The method identifies a candidate group of effective compounds through artificial intelligence based on existing compound data. Using this data, the AI system can derive a new type of compound having properties related to the target effective molecule and can simulate it. In simple words, this AI-based molecular structure design method is automatically designing the molecular structure of a new drug candidate by selectively modifying only a part of a given material structure by using the existing compound data.

Where is the startup receiving its funding from?

In July 2021, Standigm secured funding of $10 M from pavilion capital to accelerate its global competitiveness. With this funding, Standigm is ready to make more trade opportunities for its AI-driven drug assets in the global market (source).

In March 2019, Standigm raised US $11.5 million in series B round funding to advance its AI-powered drug pipelines toward license-out. This investment has participants from Kakao Ventures, Atinum Investment, DSC Investment, LB Investment, Wonik Investment Partners, as well as Mirae Asset Venture Investment and Mirae Asset Capital. Kakao Ventures, one of the leading early-stage VCs in Asia, has continued to invest in Standigm since its seed round (Source).

Collaborations

Since July 2019, Standigm and SK Chemicals Co., Ltd. are in an innovation partnership, and from then, they are working hand-in-hand in the drug-discovery process. Recently, this combo announced that they have successfully found a new rheumatoid arthritis indication for an FDA-approved drug and have filed a patent (Source).

The research collaboration is aimed at identifying novel lead compounds and repurposing existing drugs for rheumatoid arthritis and nonalcoholic steatohepatitis, leveraging Standigm’s AI-powered drug discovery platforms: Standigm BEST™, Standigm Insight™, and Standigm ASK™. In this, SK Chemicals has shared their expertise in these diseases and validated the predicted targets and compounds through in vitro and in vivo studies.

Further, in 2020, Standigm Inc. along with SK Holdings C&C, Co., opened their AI-based target identification platform, iCLUE&ASK™, on a trial basis to the public. The platform offers to prioritize protein targets for a query disease and provides the results with evidence through an interactive user interface (source).

In 2020, India-based Excelra, a global data science, and data analytics company, announced its collaboration with Standigm Inc. In this collaboration, Excelra will provide its small molecule medicinal chemistry intelligence platform GOSTAR to Standigm Inc. GOSTAR provides comprehensive information encompassing over 8 million compounds, manually curated from 3 million patents and 200,000 journal articles. The database contains over 28 million SAR-associated data points. A well-structured relational database can be utilized for diverse applications across different stages of the drug discovery and development lifecycle and aids in target validation, hit identification, early lead identification, and optimization (source).

In 2017, CrystalGenomics, Inc. and Standigm, Inc., announced their collaboration to apply Artificial Intelligence (AI) for the research and development of novel drugs. In this agreement, both parties plan to work together by combining the power of Standigm’s AI technology with CrystalGenomics’ pharmaceutical expertise to discover and develop novel drugs in the therapeutic area of cancer, rheumatoid arthritis, and liver-related diseases (source).

Genesis Therapeutics

Genesis Therapeutics is a USA-based start-up, founded in 2019. It unifies AI and biotech to accelerate the discovery of new medicines. The company uses neural networks, biophysical simulation, and a scalable computing platform for the design and development of drugs.

Usually, deep learning software just represents molecules like images and classifies them — like, say, this is a cat picture or this is not a cat picture.

But, the AI software of genesis therapeutics represents the molecules more naturally. A set of nodes or vertices, atoms, and things that connect them, bonds. They don’t just represent them as a bond or no bond. But as multiple contact types between atoms, spatial distances, and more complex features.

The resulting representation is richer and more complex. A more complete picture of a molecule than you’d get from its chemical formula, or a stick diagram showing the different structures and bonds. Because in the world of biochemistry, nothing is as simple as a diagram. Every molecule exists as a complicated, shifting 3D shape or conformation. Where important aspects like the distance between two carbon formations or bonding sites are subject to many factors. Genesis attempts to model as many of those factors as it can.

Representation is the first step; the next question is, how does one leverage that representation to learn a function that takes an input and outputs a number. Like binding affinity or solubility, or a vector that predicts multiple properties at once?

The startup is working at the intersection of modern deep neural network approaches and biophysical simulation — conformational changes in ligands and proteins.

Their Dynamic PotentialNet technology helps in protein structure prediction. It leverages 3D structural information of proteins, computational protein folding. Also, their AI platform helps in understanding the absorption, distribution, metabolism, elimination, and toxicity (ADMET) properties of drug candidates. (Source)

Source: PotentialNet for Molecular Property Prediction

So far the startup has only one patent. Nevertheless, an interesting one.

US20020046054A1 – This patent majorly focuses on a method for identifying individuals for clinical trials. This method comprises identifying and recruiting donors whose demographic characteristics, genomic and proteomic profile, and medical histories make them attractive candidates for clinical trials, drug target identification, and pharmacogenomic studies.

All in all this method enables efficient identification of research subjects and hence can effectively allow the biopharmaceutical industry to gain access to a large and varied population of individuals with detailed and fully consented medical history/data as subjects for clinical trials required for drug development and as sources of research materials.

Where is the startup receiving its funding from?

In December 2020, Genesis Therapeutics secured a $52M Series A to further accelerate AI innovation and launch a drug discovery & development pipeline. (Rock Springs Capital, T. Rowe Price Associates, Inc., Seed-round lead investor Andreessen Horowitz, Menlo Ventures, and Radical Ventures – source).

In November 2019, Genesis Therapeutics, Unifying AI and Biotech  Raised $4.1M in Seed Funding Led by Andreessen Horowitz to Accelerate and Optimize Drug Discovery/Development. (source).

Collaborations

In 2020, Genesis Therapeutics entered into a multi-target collaboration agreement with Genentech, a member of the Roche Group. The collaboration leverages Genesis’ graph machine learning and drug discovery expertise to identify innovative drug candidates for therapeutic targets in multiple disease areas (source).

Data2Discovery

USA-based Data2Discovery was founded in 2012. It is applying AI to find hidden connections and new insights in diverse, linked datasets. Their healthcare data analytics platform is intended to get results from large amounts of complex heterogeneous data. The company’s platform uses an advanced stack of scalable graph technologies, public and proprietary data sources, AI and machine learning, graph mining capabilities, and extensive experience in linking and mapping data to address the problems, enabling users to get results efficiently and effectively.

This startup has only one patent, US20190130290A1. This patent discloses a method for semantic analysis of disparate (different or diverse) data in an environment having a plurality of datasets with distinct information fields.

Further, this method involves creating graphs relating to specified information with information fields from multiple datasets as nodes with improved accuracy of machine learning in digital computers.

This technology is basically a data-driven process. There are vast data available related to drug discovery processes which include exploration data of different chemical moieties, clinical trials data of different drugs, early-stage chemical molecules, and their effectiveness data, etc.

All this data comprises some effective drug molecules which can be further explored and can become a potential drug substance for treating any condition. Data2Discovery is utilizing its platform and software to explore and screen all the available data related to the specific drug discovery processes and identify the required molecule or the step where researchers need to spend more time to get the required chemical moiety. Their platform uses all this data along with AI/ML technology to plot different scalable graphs, graph mining, and linking & mapping data to address the problems to provide the required results efficiently and effectively.

Where is the startup receiving its funding from?

In March 2017, Data2Discovery was awarded a $750,000 grant from the National Science Foundation (NSF) via the highly competitive Small Business Innovation Research (SBIR) Phase II program.

The grant will support Data2Discovery’s efforts to support translational and phenotypic research on vast interlinked datasets; which will include applications in Drug Repurposing, Toxicology and Safety, and Phenotypic Analysis (source).

Collaborations

In April 2017, the Open PHACTS Foundation announced the collaboration with Data2Discovery to form a Strategic Partnership. The goals of the partnership will support and continue with the Open PHACTS vision of creating a sustainable, open, interoperable information infrastructure for applied life science research and development while advancing science for the public benefit through shared knowledge and data in life science and biomedical research.

Data2Discovery Inc claims to bring extensive experience in pharmaceutical semantic linked data. The startup partners with pharmaceutical companies to develop full-stack semantic and graph capabilities to venture into real scientific problems (source).

In February 2017, Data2Discovery Inc identified 14 potential drug repurposing opportunities for Tuberculosis (TB) by applying its P3 graph-based association finding approach. This is done in an innovative partnership with the NIH National Center for Advancing Translational Sciences (NCATS), and OpenPHACTS. This small-scale project successfully demonstrated the feasibility of combining two key data resources – EU OpenPHACTS Open Pharmacological Space (OPS) and NCATS Phenotypic Drug Discovery Resource (PDDR) – with state-of-art graph mining tools from Data2Discovery. The capabilities demonstrated in this project open up many opportunities for public impact in rare and neglected diseases, as well as complex disease areas being pursued by pharmaceutical companies (source).

In May 2021, Data2Discovery along with Indiana University Crisis Technologies Innovation Lab and two partner companies Disaster Tech and OPS have been awarded a $2.3m contract from the US Army Telemedicine and Advanced Technology Research Center (TATRC) to create a Technology in Disaster Environments Learning Accelerator (TLA). The TLA will employ advanced data and performance science tools to identify best practices for patient care in disaster and infrastructure-degraded environments as part of the National Emergency Tele-Critical Care Network (NETCCN). Data2Discovery will use its proprietary graph technology stack along with deep expertise working with medical and biomedical data to pilot capabilities that will permit insights to be gained from multiple data streams that cannot be found elsewhere (source).

Unlearn.AI

Founded in 2017, Unlearn.AI is a platform designed to make computational clinical trials. The company’s platform accelerates clinical trials by supplementing control groups with synthetic patient data generated using AI, which helps in reducing the time to develop new medicines, enabling healthcare companies to sooner provide patients in need with life-saving therapies.

Technology – Unlearn is the only company using AI to create Digital Twins, which helps in accelerating clinical trials and getting better results. Unlearn’s platform utilizes historical datasets and disease-specific machine-learning models to generate virtual placebo patients, created from actual patient baseline data in clinical studies. This novel approach increases trial power and confidence, accelerates trial timelines, and enables patient-level insights. The whole process involves the following steps –

Step 01 [Creating a dataset]

The first step involves the preparation of a highly curated dataset, so the machine learning model can learn from the relationships.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 02 [DiGenesis™: Generating our Machine Learning Model]

After preparing the dataset, it needs to be separated into two groups, one for training, and the other for testing. The machine learning model builds an internal network of connections and starts generating Digital Twins.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 03 [Digital twin & PROCOVA™]

Once the trial has started, the platform makes records of the Digital Twin of patients. This model uses baseline data to create a complete record that predicts how the patient would have responded if he/she had not received the experimental treatment. Next, the use of PROCOVA™ (prognostic covariate adjustment) – a statistical method that incorporates Digital Twins into statistical analysis plans to provide a more precise estimate of the treatment effect.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 04 [Randomized controlled trials using Digital Twins and PROCOVA™]

The prepared digital twins are incorporated into clinical trials, which helps in enabling smaller, more efficient trials. Each patient from the trial was paired with their AI-generated predicted placebo outcome or with Digital Twin. Digital Twins maintain randomization and blinding, and increase certainty without introducing bias.

The startup has filed three patent applications. Let’s discuss each of them in brief.

  1. CA3088204A1 – This patent describes a method to train an artificial intelligence system or an artificial neural network that can provide a probability of results based on provided inputs/data. This trained AI model probability identifier can be utilized in various fields such as health informatics, image/audio processing, marketing, sociology, and lab research fields.
  2. WO2021041128A1 – This patent discloses a method for determining the treatment effects of randomized control trials (RCT). The method includes steps for receiving data from an RCT, generating results using different models (e.g., Conditional Restricted Boltzmann Machine, a recurrent neural network, etc.), and determining treatment effects for the RCT using the generated results. This method estimates quantities with high accuracy and precision and determines decision rules for declaring treatments to be effective that have low error rates.
  3. US20210117842A1 – This patent describes a method to train generative models (a machine learning model that learns to sample with the observed data) using summary statistics, so that data generated by the model satisfy specified, population-level summary statistics. Further, these generative models are utilized in a variety of fields, such as economic forecasting, climate modeling, and medical research.

All these patents explain technologies that can be used in speeding up the highly time-consuming process of clinical trials. Unlearn’s patents disclose the use of machine learning or artificial neural networks such as the restricted Boltzmann machine (RBM), a recurrent neural network, etc. in the clinical trial process and its data to get effective results.

Further, their technology also uses machine learning to create Digital Twins, which helps in accelerating clinical trials and checking the effect of particular drugs before the actual testing on humans. This technology is a breakthrough in drug discovery as it can simplify the whole clinical trial process and will provide its outcome early and accurately. Performing clinical trials using such an advanced solution will definitely reduce the time and money spent in the drug discovery process.

Also, for their breakthrough innovation using AI, Unlearn.AI won the “Predictive Analytics Solution of the Year” in the 2021 BioTech Breakthrough Awards Program (Source).

Where is the startup receiving its funding from?

In Nov. 2020, Unlearn.AI announced a series of extensions with new investments from Epic Ventures, alumni ventures groups, and global pharma company Eisai (source).

In Apr. 2020, Unlearn.AI closed a $12m series to advance the use of digital twins in clinical trials. Led by 8VC, this financing will accelerate the application of Unlearn’s innovative machine learning technology to improve clinical trial efficiency and increase confidence in results (source).

These were the 5 AI based-startups that are making ripples in Drug research. But there are more startups bringing in other technologies than AI in their drug research market.

What other technologies are companies bringing into drug research?

Apart from these technological advancements, other companies are also using different technologies in speeding up the drug discovery process. For example, an Australian-German start-up, Quantum Brilliance is using ultra-efficient quantum computers to reveal previously unknown compounds and uses quantum accelerators in the drug discovery process. Big pharma companies such as Boehringer Ingelheim,and Merck are collaborating with this start-up to expedite the drug discovery process (source).

Also, Secondcell Bio and Alliance Care Technologies International announced a new strategic partnership to use Chromovert® Technology in the rapid drug discovery process to combat rare genetic diseases. This technology uses a cell-based discovery platform to screen the cells and choose the best cell which gives effective responses to the molecules which are in the drug discovery process (source).

Future Outlook

Artificial intelligence (AI) or Machine Learning (ML) is an advanced technology that is going to hold the future in its hands. Using these technologies in the drug-discovery process will help in quickly identifying new targets and providing the cure for incurable diseases.

With the predicted market growth estimation and the increased number of working companies and startups in this domain, it can be said that the future of drug discovery seems tremendously good. But the market is still new and for drug giants to have an edge in the industry. It is important they figure out startups that align with their goals and collaborate with them.

Do you want us to help you find a startup that meets your needs?

Get in touch

Authored By: Ganesh Solanke and Nikhil Gupta, Search Team.

Here are the Top 5 AI Drug Discovery Startups:

Exscientia

 

Founded in 2012, Exscientia developed an AI-powered design platform known as Centaur Chemist, which not only identifies potential new drug targets but also builds those drugs and sends them to clinical trials.

 

 

Image Source – https://www.exscientia.ai/precision-design

With around 14 patents filed, they’ve got a considerable IP portfolio. 7 out of these 14 patents are related to the AI domain itself.

One of its recent patents – US20200013486A1 talks about an in-silico system that helps solve the real problems of the industry, which is lead optimization and identification of drug effectiveness. Now, if I take the first step in the drug discovery process- to select a chemical moiety and optimize its effectiveness, performing this task manually can be a tedious and lengthy process. But this patent proves an innovative way of lead molecule optimization via –

 

  • True lead optimization, by assessing large numbers of hypothetical molecules and converging on a more limited number of potentially active molecules, thereby reducing the number of compounds that must be synthesized and tested during a project.
  • Prediction of drug candidates having desired biological properties and activities.

It simplifies the de novo drug design process, and the costs of lead optimization are reduced, thereby having a direct impact on drug discovery.

The company has recently announced its first AI-designed molecule for treating cancer using the body’s own immune system, co-developed with Evotec. It will be entering human clinical trials soon. This happened following its 2020 announcement of bringing an AI-designed molecule for treating obsessive-compulsive disorder (OCD) to a phase 1 clinical trial in partnership with Japanese collaborator Sumitomo Dainippon Pharma.

Acquisition/Mergers and Collaborations

Exscientia has acquired Allcyte, an Austrian company developing an AI platform to predict how well cancer treatments will work in individual patients. Allcyte’s technology relies on deep learning to analyze the effects of various drugs on live samples of an individual patient’s tissue, rather than on artificial or animal models. By using allcyte’s platforms, Exscientia will be able to take a precision medicine approach to design drug molecules, ensuring that they’re even more effective at targeting tumor tissues than those designed with Exscientia’s technology alone.

Recently, the company also entered into a joint venture with GT Apeiron Therapeutics (Apeiron), a Shanghai-based company, to provide its AI-driven drug identification and design capabilities to accelerate the discovery of a therapeutic drug.

EQRx and Exscientia also team up on a discovery-through-commercialization to bring cheaper medicines to patients faster. As part of the deal, Exscientia will pick up discovery duties while EQRx will handle development and commercialization.

Where is the startup receiving its funding from?

In March 2021, Bill Gates-backed Oxford biotech raised $100M funding to discover new drugs using AI (source).

In April 2021, Softbank led a $525m round in British drug developer Exscientia (source).

Standigm

 

South Korean start-up, Standigm, founded in 2015, discovers drugs using AI, saving time and cost when compared to traditional methods. Their AI platform, Standigm BEST, is intended to explore latent chemical space to generate novel compounds with desired properties. Other than chemical data, they also analyze biomedical literature to speed up de novo drug design.

Once candidates have been identified, Standigm Insights provides biological interpretations to discover pathways, and therapeutics patterns and prioritize potential targets. The startup’s solutions eliminate uncertainty in the drug discovery process to save time and costs during development.

 

 Image credit – Standigm

In a nutshell, Standigm has achieved the automation of the whole drug discovery process based on their AI platforms, including Standigm ASK™ for target discovery, Standigm BEST™ for lead design, and Standigm Insight™ for drug repurposing. To date, Standigm has run 22 in-house or collaborative drug pipelines using ‌workflow AI technology.

The startup has filed a total of 10 patents. Given below are some of the interesting patents related to their innovative technology –

  1. KR2018022537A – This patent is mainly focusing on a method that utilizes machine learning in predicting the effects of the drug combination. This method uses a computer algorithm to collect various data –

  • cell-related data
  • drug-related data and,
  • drug/cell correlation-related data

All this collected data enables the prediction of the therapeutic efficacy of a new combination drug when synergic effects are expected according to the drug combinations. This method helps in finding out which targets or proteins are affected by which drugs. And how they interact if used in combination for the treatment purpose.

  1. KR2020145835A – This patent talks about a thioridazine composition used to treat or prevent metabolic liver disease, steatosis, inflammation, reduces fibrosis, non-alcoholic fatty liver disease, non-alcoholic steatohepatitis, etc. Further, the thioridazine molecule is selected through a screening method using artificial intelligence (AI) deep learning technology after confirming its therapeutic efficacy.
  2. WO2021080295A1 – This patent describes a method and apparatus for generating an effective compound structure by using the learning data of existing biologically active structures. The method identifies a candidate group of effective compounds through artificial intelligence based on existing compound data. Using this data, the AI system can derive a new type of compound having properties related to the target effective molecule and can simulate it. In simple words, this AI-based molecular structure design method is automatically designing the molecular structure of a new drug candidate by selectively modifying only a part of a given material structure by using the existing compound data.

Where is the startup receiving its funding from?

In July 2021, Standigm secured funding of $10 M from pavilion capital to accelerate its global competitiveness. With this funding, Standigm is ready to make more trade opportunities for its AI-driven drug assets in the global market (source).

In March 2019, Standigm raised US $11.5 million in series B round funding to advance its AI-powered drug pipelines toward license-out. This investment has participants from Kakao Ventures, Atinum Investment, DSC Investment, LB Investment, Wonik Investment Partners, as well as Mirae Asset Venture Investment and Mirae Asset Capital. Kakao Ventures, one of the leading early-stage VCs in Asia, has continued to invest in Standigm since its seed round (Source).

Collaborations

Since July 2019, Standigm and SK Chemicals Co., Ltd. are in an innovation partnership, and from then, they are working hand-in-hand in the drug-discovery process. Recently, this combo announced that they have successfully found a new rheumatoid arthritis indication for an FDA-approved drug and have filed a patent (Source).

The research collaboration is aimed at identifying novel lead compounds and repurposing existing drugs for rheumatoid arthritis and nonalcoholic steatohepatitis, leveraging Standigm’s AI-powered drug discovery platforms: Standigm BEST™, Standigm Insight™, and Standigm ASK™. In this, SK Chemicals has shared their expertise in these diseases and validated the predicted targets and compounds through in vitro and in vivo studies.

Further, in 2020, Standigm Inc. along with SK Holdings C&C, Co., opened their AI-based target identification platform, iCLUE&ASK™, on a trial basis to the public. The platform offers to prioritize protein targets for a query disease and provides the results with evidence through an interactive user interface (source).

In 2020, India-based Excelra, a global data science, and data analytics company, announced its collaboration with Standigm Inc. In this collaboration, Excelra will provide its small molecule medicinal chemistry intelligence platform GOSTAR to Standigm Inc. GOSTAR provides comprehensive information encompassing over 8 million compounds, manually curated from 3 million patents and 200,000 journal articles. The database contains over 28 million SAR-associated data points. A well-structured relational database can be utilized for diverse applications across different stages of the drug discovery and development lifecycle and aids in target validation, hit identification, early lead identification, and optimization (source).

In 2017, CrystalGenomics, Inc. and Standigm, Inc., announced their collaboration to apply Artificial Intelligence (AI) for the research and development of novel drugs. In this agreement, both parties plan to work together by combining the power of Standigm’s AI technology with CrystalGenomics’ pharmaceutical expertise to discover and develop novel drugs in the therapeutic area of cancer, rheumatoid arthritis, and liver-related diseases (source).

Genesis Therapeutics

Genesis Therapeutics is a USA-based start-up, founded in 2019. It unifies AI and biotech to accelerate the discovery of new medicines. The company uses neural networks, biophysical simulation, and a scalable computing platform for the design and development of drugs.

Usually, deep learning software just represents molecules like images and classifies them — like, say, this is a cat picture or this is not a cat picture.

But, the AI software of genesis therapeutics represents the molecules more naturally. A set of nodes or vertices, atoms, and things that connect them, bonds. They don’t just represent them as a bond or no bond. But as multiple contact types between atoms, spatial distances, and more complex features.

The resulting representation is richer and more complex. A more complete picture of a molecule than you’d get from its chemical formula, or a stick diagram showing the different structures and bonds. Because in the world of biochemistry, nothing is as simple as a diagram. Every molecule exists as a complicated, shifting 3D shape or conformation. Where important aspects like the distance between two carbon formations or bonding sites are subject to many factors. Genesis attempts to model as many of those factors as it can.

Representation is the first step; the next question is, how does one leverage that representation to learn a function that takes an input and outputs a number. Like binding affinity or solubility, or a vector that predicts multiple properties at once?

The startup is working at the intersection of modern deep neural network approaches and biophysical simulation — conformational changes in ligands and proteins.

Their Dynamic PotentialNet technology helps in protein structure prediction. It leverages 3D structural information of proteins, computational protein folding. Also, their AI platform helps in understanding the absorption, distribution, metabolism, elimination, and toxicity (ADMET) properties of drug candidates. (Source)

Source: PotentialNet for Molecular Property Prediction

So far the startup has only one patent. Nevertheless, an interesting one.

US20020046054A1 – This patent majorly focuses on a method for identifying individuals for clinical trials. This method comprises identifying and recruiting donors whose demographic characteristics, genomic and proteomic profile, and medical histories make them attractive candidates for clinical trials, drug target identification, and pharmacogenomic studies.

All in all this method enables efficient identification of research subjects and hence can effectively allow the biopharmaceutical industry to gain access to a large and varied population of individuals with detailed and fully consented medical history/data as subjects for clinical trials required for drug development and as sources of research materials.

Where is the startup receiving its funding from?

In December 2020, Genesis Therapeutics secured a $52M Series A to further accelerate AI innovation and launch a drug discovery & development pipeline. (Rock Springs Capital, T. Rowe Price Associates, Inc., Seed-round lead investor Andreessen Horowitz, Menlo Ventures, and Radical Ventures – source).

In November 2019, Genesis Therapeutics, Unifying AI and Biotech  Raised $4.1M in Seed Funding Led by Andreessen Horowitz to Accelerate and Optimize Drug Discovery/Development. (source).

Collaborations

In 2020, Genesis Therapeutics entered into a multi-target collaboration agreement with Genentech, a member of the Roche Group. The collaboration leverages Genesis’ graph machine learning and drug discovery expertise to identify innovative drug candidates for therapeutic targets in multiple disease areas (source).

Data2Discovery

USA-based Data2Discovery was founded in 2012. It is applying AI to find hidden connections and new insights in diverse, linked datasets. Their healthcare data analytics platform is intended to get results from large amounts of complex heterogeneous data. The company’s platform uses an advanced stack of scalable graph technologies, public and proprietary data sources, AI and machine learning, graph mining capabilities, and extensive experience in linking and mapping data to address the problems, enabling users to get results efficiently and effectively.

This startup has only one patent, US20190130290A1. This patent discloses a method for semantic analysis of disparate (different or diverse) data in an environment having a plurality of datasets with distinct information fields.

Further, this method involves creating graphs relating to specified information with information fields from multiple datasets as nodes with improved accuracy of machine learning in digital computers.

This technology is basically a data-driven process. There are vast data available related to drug discovery processes which include exploration data of different chemical moieties, clinical trials data of different drugs, early-stage chemical molecules, and their effectiveness data, etc.

All this data comprises some effective drug molecules which can be further explored and can become a potential drug substance for treating any condition. Data2Discovery is utilizing its platform and software to explore and screen all the available data related to the specific drug discovery processes and identify the required molecule or the step where researchers need to spend more time to get the required chemical moiety. Their platform uses all this data along with AI/ML technology to plot different scalable graphs, graph mining, and linking & mapping data to address the problems to provide the required results efficiently and effectively.

Where is the startup receiving its funding from?

In March 2017, Data2Discovery was awarded a $750,000 grant from the National Science Foundation (NSF) via the highly competitive Small Business Innovation Research (SBIR) Phase II program.

The grant will support Data2Discovery’s efforts to support translational and phenotypic research on vast interlinked datasets; which will include applications in Drug Repurposing, Toxicology and Safety, and Phenotypic Analysis (source).

Collaborations

In April 2017, the Open PHACTS Foundation announced the collaboration with Data2Discovery to form a Strategic Partnership. The goals of the partnership will support and continue with the Open PHACTS vision of creating a sustainable, open, interoperable information infrastructure for applied life science research and development while advancing science for the public benefit through shared knowledge and data in life science and biomedical research.

Data2Discovery Inc claims to bring extensive experience in pharmaceutical semantic linked data. The startup partners with pharmaceutical companies to develop full-stack semantic and graph capabilities to venture into real scientific problems (source).

In February 2017, Data2Discovery Inc identified 14 potential drug repurposing opportunities for Tuberculosis (TB) by applying its P3 graph-based association finding approach. This is done in an innovative partnership with the NIH National Center for Advancing Translational Sciences (NCATS), and OpenPHACTS. This small-scale project successfully demonstrated the feasibility of combining two key data resources – EU OpenPHACTS Open Pharmacological Space (OPS) and NCATS Phenotypic Drug Discovery Resource (PDDR) – with state-of-art graph mining tools from Data2Discovery. The capabilities demonstrated in this project open up many opportunities for public impact in rare and neglected diseases, as well as complex disease areas being pursued by pharmaceutical companies (source).

In May 2021, Data2Discovery along with Indiana University Crisis Technologies Innovation Lab and two partner companies Disaster Tech and OPS have been awarded a $2.3m contract from the US Army Telemedicine and Advanced Technology Research Center (TATRC) to create a Technology in Disaster Environments Learning Accelerator (TLA). The TLA will employ advanced data and performance science tools to identify best practices for patient care in disaster and infrastructure-degraded environments as part of the National Emergency Tele-Critical Care Network (NETCCN). Data2Discovery will use its proprietary graph technology stack along with deep expertise working with medical and biomedical data to pilot capabilities that will permit insights to be gained from multiple data streams that cannot be found elsewhere (source).

Unlearn.AI

Founded in 2017, Unlearn.AI is a platform designed to make computational clinical trials. The company’s platform accelerates clinical trials by supplementing control groups with synthetic patient data generated using AI, which helps in reducing the time to develop new medicines, enabling healthcare companies to sooner provide patients in need with life-saving therapies.

Technology – Unlearn is the only company using AI to create Digital Twins, which helps in accelerating clinical trials and getting better results. Unlearn’s platform utilizes historical datasets and disease-specific machine-learning models to generate virtual placebo patients, created from actual patient baseline data in clinical studies. This novel approach increases trial power and confidence, accelerates trial timelines, and enables patient-level insights. The whole process involves the following steps –

Step 01 [Creating a dataset]

The first step involves the preparation of a highly curated dataset, so the machine learning model can learn from the relationships.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 02 [DiGenesis™: Generating our Machine Learning Model]

After preparing the dataset, it needs to be separated into two groups, one for training, and the other for testing. The machine learning model builds an internal network of connections and starts generating Digital Twins.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 03 [Digital twin & PROCOVA™]

Once the trial has started, the platform makes records of the Digital Twin of patients. This model uses baseline data to create a complete record that predicts how the patient would have responded if he/she had not received the experimental treatment. Next, the use of PROCOVA™ (prognostic covariate adjustment) – a statistical method that incorporates Digital Twins into statistical analysis plans to provide a more precise estimate of the treatment effect.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 04 [Randomized controlled trials using Digital Twins and PROCOVA™]

The prepared digital twins are incorporated into clinical trials, which helps in enabling smaller, more efficient trials. Each patient from the trial was paired with their AI-generated predicted placebo outcome or with Digital Twin. Digital Twins maintain randomization and blinding, and increase certainty without introducing bias.

The startup has filed three patent applications. Let’s discuss each of them in brief.

  1. CA3088204A1 – This patent describes a method to train an artificial intelligence system or an artificial neural network that can provide a probability of results based on provided inputs/data. This trained AI model probability identifier can be utilized in various fields such as health informatics, image/audio processing, marketing, sociology, and lab research fields.
  2. WO2021041128A1 – This patent discloses a method for determining the treatment effects of randomized control trials (RCT). The method includes steps for receiving data from an RCT, generating results using different models (e.g., Conditional Restricted Boltzmann Machine, a recurrent neural network, etc.), and determining treatment effects for the RCT using the generated results. This method estimates quantities with high accuracy and precision and determines decision rules for declaring treatments to be effective that have low error rates.
  3. US20210117842A1 – This patent describes a method to train generative models (a machine learning model that learns to sample with the observed data) using summary statistics, so that data generated by the model satisfy specified, population-level summary statistics. Further, these generative models are utilized in a variety of fields, such as economic forecasting, climate modeling, and medical research.

All these patents explain technologies that can be used in speeding up the highly time-consuming process of clinical trials. Unlearn’s patents disclose the use of machine learning or artificial neural networks such as the restricted Boltzmann machine (RBM), a recurrent neural network, etc. in the clinical trial process and its data to get effective results.

Further, their technology also uses machine learning to create Digital Twins, which helps in accelerating clinical trials and checking the effect of particular drugs before the actual testing on humans. This technology is a breakthrough in drug discovery as it can simplify the whole clinical trial process and will provide its outcome early and accurately. Performing clinical trials using such an advanced solution will definitely reduce the time and money spent in the drug discovery process.

Also, for their breakthrough innovation using AI, Unlearn.AI won the “Predictive Analytics Solution of the Year” in the 2021 BioTech Breakthrough Awards Program (Source).

Where is the startup receiving its funding from?

In Nov. 2020, Unlearn.AI announced a series of extensions with new investments from Epic Ventures, alumni ventures groups, and global pharma company Eisai (source).

In Apr. 2020, Unlearn.AI closed a $12m series to advance the use of digital twins in clinical trials. Led by 8VC, this financing will accelerate the application of Unlearn’s innovative machine learning technology to improve clinical trial efficiency and increase confidence in results (source).

These were the 5 AI based-startups that are making ripples in Drug research. But there are more startups bringing in other technologies than AI in their drug research market.

What other technologies are companies bringing into drug research?

Apart from these technological advancements, other companies are also using different technologies in speeding up the drug discovery process. For example, an Australian-German start-up, Quantum Brilliance is using ultra-efficient quantum computers to reveal previously unknown compounds and uses quantum accelerators in the drug discovery process. Big pharma companies such as Boehringer Ingelheim,and Merck are collaborating with this start-up to expedite the drug discovery process (source).

Also, Secondcell Bio and Alliance Care Technologies International announced a new strategic partnership to use Chromovert® Technology in the rapid drug discovery process to combat rare genetic diseases. This technology uses a cell-based discovery platform to screen the cells and choose the best cell which gives effective responses to the molecules which are in the drug discovery process (source).

Future Outlook

Artificial intelligence (AI) or Machine Learning (ML) is an advanced technology that is going to hold the future in its hands. Using these technologies in the drug-discovery process will help in quickly identifying new targets and providing the cure for incurable diseases.

With the predicted market growth estimation and the increased number of working companies and startups in this domain, it can be said that the future of drug discovery seems tremendously good. But the market is still new and for drug giants to have an edge in the industry. It is important they figure out startups that align with their goals and collaborate with them.

Do you want us to help you find a startup that meets your needs?

Get in touch

Authored By: Ganesh Solanke and Nikhil Gupta, Search Team.

To check different players from this domain, we first prepared targeted search queries which we executed on patent databases. From there, we observed that a few companies have patents on ‌drug discovery processes using advanced technologies like AI, ML, etc. In the next step, we checked these companies’ other market activities such as-

  1. Funding amount received from Government, other companies, etc.
  2. Collaboration/partnership for developing new solutions or licensing their technologies to other future leaders from this domain
  3. Merger and acquisition with other same domain companies

After checking all the patent and market data, we narrowed the list to the following 5 start-ups. These startups are not just offering unique solutions to the market but are actively filing patents for their innovations.

Here are the Top 5 AI Drug Discovery Startups:

Exscientia

 

Founded in 2012, Exscientia developed an AI-powered design platform known as Centaur Chemist, which not only identifies potential new drug targets but also builds those drugs and sends them to clinical trials.

 

 

Image Source – https://www.exscientia.ai/precision-design

With around 14 patents filed, they’ve got a considerable IP portfolio. 7 out of these 14 patents are related to the AI domain itself.

One of its recent patents – US20200013486A1 talks about an in-silico system that helps solve the real problems of the industry, which is lead optimization and identification of drug effectiveness. Now, if I take the first step in the drug discovery process- to select a chemical moiety and optimize its effectiveness, performing this task manually can be a tedious and lengthy process. But this patent proves an innovative way of lead molecule optimization via –

 

  • True lead optimization, by assessing large numbers of hypothetical molecules and converging on a more limited number of potentially active molecules, thereby reducing the number of compounds that must be synthesized and tested during a project.
  • Prediction of drug candidates having desired biological properties and activities.

It simplifies the de novo drug design process, and the costs of lead optimization are reduced, thereby having a direct impact on drug discovery.

The company has recently announced its first AI-designed molecule for treating cancer using the body’s own immune system, co-developed with Evotec. It will be entering human clinical trials soon. This happened following its 2020 announcement of bringing an AI-designed molecule for treating obsessive-compulsive disorder (OCD) to a phase 1 clinical trial in partnership with Japanese collaborator Sumitomo Dainippon Pharma.

Acquisition/Mergers and Collaborations

Exscientia has acquired Allcyte, an Austrian company developing an AI platform to predict how well cancer treatments will work in individual patients. Allcyte’s technology relies on deep learning to analyze the effects of various drugs on live samples of an individual patient’s tissue, rather than on artificial or animal models. By using allcyte’s platforms, Exscientia will be able to take a precision medicine approach to design drug molecules, ensuring that they’re even more effective at targeting tumor tissues than those designed with Exscientia’s technology alone.

Recently, the company also entered into a joint venture with GT Apeiron Therapeutics (Apeiron), a Shanghai-based company, to provide its AI-driven drug identification and design capabilities to accelerate the discovery of a therapeutic drug.

EQRx and Exscientia also team up on a discovery-through-commercialization to bring cheaper medicines to patients faster. As part of the deal, Exscientia will pick up discovery duties while EQRx will handle development and commercialization.

Where is the startup receiving its funding from?

In March 2021, Bill Gates-backed Oxford biotech raised $100M funding to discover new drugs using AI (source).

In April 2021, Softbank led a $525m round in British drug developer Exscientia (source).

Standigm

 

South Korean start-up, Standigm, founded in 2015, discovers drugs using AI, saving time and cost when compared to traditional methods. Their AI platform, Standigm BEST, is intended to explore latent chemical space to generate novel compounds with desired properties. Other than chemical data, they also analyze biomedical literature to speed up de novo drug design.

Once candidates have been identified, Standigm Insights provides biological interpretations to discover pathways, and therapeutics patterns and prioritize potential targets. The startup’s solutions eliminate uncertainty in the drug discovery process to save time and costs during development.

 

 Image credit – Standigm

In a nutshell, Standigm has achieved the automation of the whole drug discovery process based on their AI platforms, including Standigm ASK™ for target discovery, Standigm BEST™ for lead design, and Standigm Insight™ for drug repurposing. To date, Standigm has run 22 in-house or collaborative drug pipelines using ‌workflow AI technology.

The startup has filed a total of 10 patents. Given below are some of the interesting patents related to their innovative technology –

  1. KR2018022537A – This patent is mainly focusing on a method that utilizes machine learning in predicting the effects of the drug combination. This method uses a computer algorithm to collect various data –

  • cell-related data
  • drug-related data and,
  • drug/cell correlation-related data

All this collected data enables the prediction of the therapeutic efficacy of a new combination drug when synergic effects are expected according to the drug combinations. This method helps in finding out which targets or proteins are affected by which drugs. And how they interact if used in combination for the treatment purpose.

  1. KR2020145835A – This patent talks about a thioridazine composition used to treat or prevent metabolic liver disease, steatosis, inflammation, reduces fibrosis, non-alcoholic fatty liver disease, non-alcoholic steatohepatitis, etc. Further, the thioridazine molecule is selected through a screening method using artificial intelligence (AI) deep learning technology after confirming its therapeutic efficacy.
  2. WO2021080295A1 – This patent describes a method and apparatus for generating an effective compound structure by using the learning data of existing biologically active structures. The method identifies a candidate group of effective compounds through artificial intelligence based on existing compound data. Using this data, the AI system can derive a new type of compound having properties related to the target effective molecule and can simulate it. In simple words, this AI-based molecular structure design method is automatically designing the molecular structure of a new drug candidate by selectively modifying only a part of a given material structure by using the existing compound data.

Where is the startup receiving its funding from?

In July 2021, Standigm secured funding of $10 M from pavilion capital to accelerate its global competitiveness. With this funding, Standigm is ready to make more trade opportunities for its AI-driven drug assets in the global market (source).

In March 2019, Standigm raised US $11.5 million in series B round funding to advance its AI-powered drug pipelines toward license-out. This investment has participants from Kakao Ventures, Atinum Investment, DSC Investment, LB Investment, Wonik Investment Partners, as well as Mirae Asset Venture Investment and Mirae Asset Capital. Kakao Ventures, one of the leading early-stage VCs in Asia, has continued to invest in Standigm since its seed round (Source).

Collaborations

Since July 2019, Standigm and SK Chemicals Co., Ltd. are in an innovation partnership, and from then, they are working hand-in-hand in the drug-discovery process. Recently, this combo announced that they have successfully found a new rheumatoid arthritis indication for an FDA-approved drug and have filed a patent (Source).

The research collaboration is aimed at identifying novel lead compounds and repurposing existing drugs for rheumatoid arthritis and nonalcoholic steatohepatitis, leveraging Standigm’s AI-powered drug discovery platforms: Standigm BEST™, Standigm Insight™, and Standigm ASK™. In this, SK Chemicals has shared their expertise in these diseases and validated the predicted targets and compounds through in vitro and in vivo studies.

Further, in 2020, Standigm Inc. along with SK Holdings C&C, Co., opened their AI-based target identification platform, iCLUE&ASK™, on a trial basis to the public. The platform offers to prioritize protein targets for a query disease and provides the results with evidence through an interactive user interface (source).

In 2020, India-based Excelra, a global data science, and data analytics company, announced its collaboration with Standigm Inc. In this collaboration, Excelra will provide its small molecule medicinal chemistry intelligence platform GOSTAR to Standigm Inc. GOSTAR provides comprehensive information encompassing over 8 million compounds, manually curated from 3 million patents and 200,000 journal articles. The database contains over 28 million SAR-associated data points. A well-structured relational database can be utilized for diverse applications across different stages of the drug discovery and development lifecycle and aids in target validation, hit identification, early lead identification, and optimization (source).

In 2017, CrystalGenomics, Inc. and Standigm, Inc., announced their collaboration to apply Artificial Intelligence (AI) for the research and development of novel drugs. In this agreement, both parties plan to work together by combining the power of Standigm’s AI technology with CrystalGenomics’ pharmaceutical expertise to discover and develop novel drugs in the therapeutic area of cancer, rheumatoid arthritis, and liver-related diseases (source).

Genesis Therapeutics

Genesis Therapeutics is a USA-based start-up, founded in 2019. It unifies AI and biotech to accelerate the discovery of new medicines. The company uses neural networks, biophysical simulation, and a scalable computing platform for the design and development of drugs.

Usually, deep learning software just represents molecules like images and classifies them — like, say, this is a cat picture or this is not a cat picture.

But, the AI software of genesis therapeutics represents the molecules more naturally. A set of nodes or vertices, atoms, and things that connect them, bonds. They don’t just represent them as a bond or no bond. But as multiple contact types between atoms, spatial distances, and more complex features.

The resulting representation is richer and more complex. A more complete picture of a molecule than you’d get from its chemical formula, or a stick diagram showing the different structures and bonds. Because in the world of biochemistry, nothing is as simple as a diagram. Every molecule exists as a complicated, shifting 3D shape or conformation. Where important aspects like the distance between two carbon formations or bonding sites are subject to many factors. Genesis attempts to model as many of those factors as it can.

Representation is the first step; the next question is, how does one leverage that representation to learn a function that takes an input and outputs a number. Like binding affinity or solubility, or a vector that predicts multiple properties at once?

The startup is working at the intersection of modern deep neural network approaches and biophysical simulation — conformational changes in ligands and proteins.

Their Dynamic PotentialNet technology helps in protein structure prediction. It leverages 3D structural information of proteins, computational protein folding. Also, their AI platform helps in understanding the absorption, distribution, metabolism, elimination, and toxicity (ADMET) properties of drug candidates. (Source)

Source: PotentialNet for Molecular Property Prediction

So far the startup has only one patent. Nevertheless, an interesting one.

US20020046054A1 – This patent majorly focuses on a method for identifying individuals for clinical trials. This method comprises identifying and recruiting donors whose demographic characteristics, genomic and proteomic profile, and medical histories make them attractive candidates for clinical trials, drug target identification, and pharmacogenomic studies.

All in all this method enables efficient identification of research subjects and hence can effectively allow the biopharmaceutical industry to gain access to a large and varied population of individuals with detailed and fully consented medical history/data as subjects for clinical trials required for drug development and as sources of research materials.

Where is the startup receiving its funding from?

In December 2020, Genesis Therapeutics secured a $52M Series A to further accelerate AI innovation and launch a drug discovery & development pipeline. (Rock Springs Capital, T. Rowe Price Associates, Inc., Seed-round lead investor Andreessen Horowitz, Menlo Ventures, and Radical Ventures – source).

In November 2019, Genesis Therapeutics, Unifying AI and Biotech  Raised $4.1M in Seed Funding Led by Andreessen Horowitz to Accelerate and Optimize Drug Discovery/Development. (source).

Collaborations

In 2020, Genesis Therapeutics entered into a multi-target collaboration agreement with Genentech, a member of the Roche Group. The collaboration leverages Genesis’ graph machine learning and drug discovery expertise to identify innovative drug candidates for therapeutic targets in multiple disease areas (source).

Data2Discovery

USA-based Data2Discovery was founded in 2012. It is applying AI to find hidden connections and new insights in diverse, linked datasets. Their healthcare data analytics platform is intended to get results from large amounts of complex heterogeneous data. The company’s platform uses an advanced stack of scalable graph technologies, public and proprietary data sources, AI and machine learning, graph mining capabilities, and extensive experience in linking and mapping data to address the problems, enabling users to get results efficiently and effectively.

This startup has only one patent, US20190130290A1. This patent discloses a method for semantic analysis of disparate (different or diverse) data in an environment having a plurality of datasets with distinct information fields.

Further, this method involves creating graphs relating to specified information with information fields from multiple datasets as nodes with improved accuracy of machine learning in digital computers.

This technology is basically a data-driven process. There are vast data available related to drug discovery processes which include exploration data of different chemical moieties, clinical trials data of different drugs, early-stage chemical molecules, and their effectiveness data, etc.

All this data comprises some effective drug molecules which can be further explored and can become a potential drug substance for treating any condition. Data2Discovery is utilizing its platform and software to explore and screen all the available data related to the specific drug discovery processes and identify the required molecule or the step where researchers need to spend more time to get the required chemical moiety. Their platform uses all this data along with AI/ML technology to plot different scalable graphs, graph mining, and linking & mapping data to address the problems to provide the required results efficiently and effectively.

Where is the startup receiving its funding from?

In March 2017, Data2Discovery was awarded a $750,000 grant from the National Science Foundation (NSF) via the highly competitive Small Business Innovation Research (SBIR) Phase II program.

The grant will support Data2Discovery’s efforts to support translational and phenotypic research on vast interlinked datasets; which will include applications in Drug Repurposing, Toxicology and Safety, and Phenotypic Analysis (source).

Collaborations

In April 2017, the Open PHACTS Foundation announced the collaboration with Data2Discovery to form a Strategic Partnership. The goals of the partnership will support and continue with the Open PHACTS vision of creating a sustainable, open, interoperable information infrastructure for applied life science research and development while advancing science for the public benefit through shared knowledge and data in life science and biomedical research.

Data2Discovery Inc claims to bring extensive experience in pharmaceutical semantic linked data. The startup partners with pharmaceutical companies to develop full-stack semantic and graph capabilities to venture into real scientific problems (source).

In February 2017, Data2Discovery Inc identified 14 potential drug repurposing opportunities for Tuberculosis (TB) by applying its P3 graph-based association finding approach. This is done in an innovative partnership with the NIH National Center for Advancing Translational Sciences (NCATS), and OpenPHACTS. This small-scale project successfully demonstrated the feasibility of combining two key data resources – EU OpenPHACTS Open Pharmacological Space (OPS) and NCATS Phenotypic Drug Discovery Resource (PDDR) – with state-of-art graph mining tools from Data2Discovery. The capabilities demonstrated in this project open up many opportunities for public impact in rare and neglected diseases, as well as complex disease areas being pursued by pharmaceutical companies (source).

In May 2021, Data2Discovery along with Indiana University Crisis Technologies Innovation Lab and two partner companies Disaster Tech and OPS have been awarded a $2.3m contract from the US Army Telemedicine and Advanced Technology Research Center (TATRC) to create a Technology in Disaster Environments Learning Accelerator (TLA). The TLA will employ advanced data and performance science tools to identify best practices for patient care in disaster and infrastructure-degraded environments as part of the National Emergency Tele-Critical Care Network (NETCCN). Data2Discovery will use its proprietary graph technology stack along with deep expertise working with medical and biomedical data to pilot capabilities that will permit insights to be gained from multiple data streams that cannot be found elsewhere (source).

Unlearn.AI

Founded in 2017, Unlearn.AI is a platform designed to make computational clinical trials. The company’s platform accelerates clinical trials by supplementing control groups with synthetic patient data generated using AI, which helps in reducing the time to develop new medicines, enabling healthcare companies to sooner provide patients in need with life-saving therapies.

Technology – Unlearn is the only company using AI to create Digital Twins, which helps in accelerating clinical trials and getting better results. Unlearn’s platform utilizes historical datasets and disease-specific machine-learning models to generate virtual placebo patients, created from actual patient baseline data in clinical studies. This novel approach increases trial power and confidence, accelerates trial timelines, and enables patient-level insights. The whole process involves the following steps –

Step 01 [Creating a dataset]

The first step involves the preparation of a highly curated dataset, so the machine learning model can learn from the relationships.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 02 [DiGenesis™: Generating our Machine Learning Model]

After preparing the dataset, it needs to be separated into two groups, one for training, and the other for testing. The machine learning model builds an internal network of connections and starts generating Digital Twins.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 03 [Digital twin & PROCOVA™]

Once the trial has started, the platform makes records of the Digital Twin of patients. This model uses baseline data to create a complete record that predicts how the patient would have responded if he/she had not received the experimental treatment. Next, the use of PROCOVA™ (prognostic covariate adjustment) – a statistical method that incorporates Digital Twins into statistical analysis plans to provide a more precise estimate of the treatment effect.

Image Source – https://www.unlearn.ai/solutions#intelligent

Step 04 [Randomized controlled trials using Digital Twins and PROCOVA™]

The prepared digital twins are incorporated into clinical trials, which helps in enabling smaller, more efficient trials. Each patient from the trial was paired with their AI-generated predicted placebo outcome or with Digital Twin. Digital Twins maintain randomization and blinding, and increase certainty without introducing bias.

The startup has filed three patent applications. Let’s discuss each of them in brief.

  1. CA3088204A1 – This patent describes a method to train an artificial intelligence system or an artificial neural network that can provide a probability of results based on provided inputs/data. This trained AI model probability identifier can be utilized in various fields such as health informatics, image/audio processing, marketing, sociology, and lab research fields.
  2. WO2021041128A1 – This patent discloses a method for determining the treatment effects of randomized control trials (RCT). The method includes steps for receiving data from an RCT, generating results using different models (e.g., Conditional Restricted Boltzmann Machine, a recurrent neural network, etc.), and determining treatment effects for the RCT using the generated results. This method estimates quantities with high accuracy and precision and determines decision rules for declaring treatments to be effective that have low error rates.
  3. US20210117842A1 – This patent describes a method to train generative models (a machine learning model that learns to sample with the observed data) using summary statistics, so that data generated by the model satisfy specified, population-level summary statistics. Further, these generative models are utilized in a variety of fields, such as economic forecasting, climate modeling, and medical research.

All these patents explain technologies that can be used in speeding up the highly time-consuming process of clinical trials. Unlearn’s patents disclose the use of machine learning or artificial neural networks such as the restricted Boltzmann machine (RBM), a recurrent neural network, etc. in the clinical trial process and its data to get effective results.

Further, their technology also uses machine learning to create Digital Twins, which helps in accelerating clinical trials and checking the effect of particular drugs before the actual testing on humans. This technology is a breakthrough in drug discovery as it can simplify the whole clinical trial process and will provide its outcome early and accurately. Performing clinical trials using such an advanced solution will definitely reduce the time and money spent in the drug discovery process.

Also, for their breakthrough innovation using AI, Unlearn.AI won the “Predictive Analytics Solution of the Year” in the 2021 BioTech Breakthrough Awards Program (Source).

Where is the startup receiving its funding from?

In Nov. 2020, Unlearn.AI announced a series of extensions with new investments from Epic Ventures, alumni ventures groups, and global pharma company Eisai (source).

In Apr. 2020, Unlearn.AI closed a $12m series to advance the use of digital twins in clinical trials. Led by 8VC, this financing will accelerate the application of Unlearn’s innovative machine learning technology to improve clinical trial efficiency and increase confidence in results (source).

These were the 5 AI based-startups that are making ripples in Drug research. But there are more startups bringing in other technologies than AI in their drug research market.

What other technologies are companies bringing into drug research?

Apart from these technological advancements, other companies are also using different technologies in speeding up the drug discovery process. For example, an Australian-German start-up, Quantum Brilliance is using ultra-efficient quantum computers to reveal previously unknown compounds and uses quantum accelerators in the drug discovery process. Big pharma companies such as Boehringer Ingelheim,and Merck are collaborating with this start-up to expedite the drug discovery process (source).

Also, Secondcell Bio and Alliance Care Technologies International announced a new strategic partnership to use Chromovert® Technology in the rapid drug discovery process to combat rare genetic diseases. This technology uses a cell-based discovery platform to screen the cells and choose the best cell which gives effective responses to the molecules which are in the drug discovery process (source).

Future Outlook

Artificial intelligence (AI) or Machine Learning (ML) is an advanced technology that is going to hold the future in its hands. Using these technologies in the drug-discovery process will help in quickly identifying new targets and providing the cure for incurable diseases.

With the predicted market growth estimation and the increased number of working companies and startups in this domain, it can be said that the future of drug discovery seems tremendously good. But the market is still new and for drug giants to have an edge in the industry. It is important they figure out startups that align with their goals and collaborate with them.

Do you want us to help you find a startup that meets your needs?

Get in touch

Authored By: Ganesh Solanke and Nikhil Gupta, Search Team.

Leave a Comment