13类97家利用AI发现新药的初创公司
Sometime ago, I wrote about how we're nowin the long-tail of machine learning indrug discovery. I noted that we'removing past generalist applications of AIsuch as IBM Watson's to more specific,purpose-built tools. This got methinking: What are all the startups applyingartificial intelligence in drugdiscovery currently?
Accordingly,I did some research anddeveloped the following list, which I have groupedroughly by research phase. Iupdate this list monthly. I'm sure that I'vemissed some, so please add any I'vemissed in the comments. Here's what I'veuncovered so far:
Aggregate and Synthesize Information
Understand Mechanisms of Disease
Generate Data and Models
Repurpose Existing Drugs
Generate Novel Drug Candidates
Validate Drug Candidates
Design Drugs
Design Preclinical Experiments
Run Preclinical Experiments
Design Clinical Trials
Recruit for Clinical Trials
Optimize Clinical Trials
Publish Data
1) Aggregate and Synthesize Information
Uses AI to: Process raw phenotypic, imaging, drug, and genomic data sets. Allows researchers to: Integrate rapid analytics and machine learning capabilities into existing business processes to improve care, enhance discoveries, gain insight into business, and enable fast data-driven decisions.
Biorelate
Uses AI to: Create curated databases from the analysis of published scientific literature. Allows researchers to: Extract structured biological knowledge to power drug discovery applications.
Causaly
Uses AI to: Read scientific articles and extract causal associations. Allows researchers to: Search for cause and effect relationships and gather evidence on how biomedical entities interact.
Uses AI to: Build a database of therapy evidence. Allows researchers to: Quickly answer any comparative cost and outcome question.
Helix
Uses AI to: Respond to verbal questions and requests in a lab setting. Allows researchers to: Increase efficiency, improve lab safety, keep current on relevant new research, and manage inventory.
Uses AI to: Generate insights from billions of disparate data points from thousands of data sources. Allows researchers to: Improve decision-making by seeing information in context from biomedical data sources including publications, clinical trials, congresses, and theses.
Uses AI to: Learn underlying correlations in fragmented datasets with incomplete information. Allows researchers to: Estimate missing knowledge of how candidate drugs act on proteins, to aid design of new drug cocktails that activate proteins to cure disease.
Uses AI to: Establish and find the similarity of document "fingerprints" based on a combination of keyword extraction, word embeddings, neural topic modeling, and other natural language understanding techniques. Allows researchers to: Build reading lists faster, with more precision and interdisciplinary inspiration.
Uses AI to: Analyze large amounts of data fast and provide explainable, auditable results. Allows researchers to: Understand and extract meaning from internal data sets, especially unstructured ones.
Uses AI to: Analyze and organize biomedical research. Allows researchers to: Receive a personalized feed of the most relevant and important research as it's published.
Mozi
Uses AI to: Find patterns in biomedical data and infer hypotheses for investigation. Allows researchers to: Upload datasets and have them analyzed in the context of global biomedical knowledge, leading to new diagnostic and treatment strategies, particularly for personalized medicine.
Uses AI to: Extract knowledge in real-time from commercial, scientific, and regulatory literature. Allows researchers to: Identify competitive white space, eliminate blind spots in research, and discover disease similarities by phenotype for clinical trial design.
Uses AI to: Build intelligence from distributed datasets, including through privacy-safe transfer and federated learning. Allows researchers to: Overcome the problem of data-sharing in healthcare to automate diagnostics, predict treatment outcomes, and optimize clinical trials.
Uses AI to: Analyze over 114 million chemical structures, clinical trial information, regulatory details, toxicity data, and over 121 million patents and other sources. Allows researchers to: Validate chemical development projects.
Uses AI to: Allow for intuitive searches on the world's biomedical research data. Allows researchers to: Find relevant results for drug discovery-related queries such as compounds for a specified target.
Uses AI to: Search and organize information in research papers, clinical trial listings, and patents. Allows researchers to: Save time searching for information while receiving more relevant and personalized results.
Uses AI to: Curate, in combination with human expertise, millions of scientific papers from thousands of publications. Allows researchers to: Stay up-to-date with new scientific publications and patents.
Uses AI to: Enable natural language search on billions of rows of data from any source. Allows researchers to: Speed analysis of clinical trial results and historical genomics data.
2) Understand Mechanisms of Disease
Uses AI to: Predict cancer progression from tumor DNA in blood samples. Allows researchers to: Determine treatment response and relapse earlier, and use Bayesian adaptive clinical trial design to increase the success of late stage trials.
Uses AI to: Organize and standardize immune-related gene, protein, cell, and microbiome data into a single, machine-readable, cell-level view of the immune-system. Allows researchers to: Gain novel insights related to mechanisms of disease, clinical markers, and drug discovery and validation.
Uses AI to: Analyze 200 omics databases, connecting published literature, experimental data, and clinical data. Allows researchers to: Get insight into how molecular mechanisms influence cell and tissue functions, and how these in turn influence phenotypes and disease pathology.
Uses AI to: Link phenotypic traits to genetic mutations. Allows researchers to: Discover new clinical signs, symptoms, and genes for biomarkers, and access data to develop, test, and market precision medicines.
Uses AI to: Analyze cell and tissue phenotypes in microscopy data. Allows researchers to: Rapidly and accurately profile single cells in microscopy images.
Uses AI to: Analyze metabolomic data along with other large-scale molecular information such as data from genes, proteins, drugs, and diseases. Allows researchers to: Find disease pathways, novel drug targets, new therapeutic effects for existing drugs, molecular mechanisms for pharmacological effects, and new biomarkers.
Uses AI to: Enable high-throughput structure discovery of proteins and molecular complexes from cryo-EM data. Allows researchers to: Discover and understand the detailed three-dimensional structure of important protein molecules, complexes, and drug targets.
3) Generate Data and Models
Uses AI to: Train machine learning models on large, high-quality datasets. Allows researchers to: Address key problems in the drug discovery and development process.
4) Repurpose Existing Drugs
Uses AI to: Analyze data from sources including patient stem cells. Allows researchers to: Discover drugs for neurodenerative diseases, including ALS.
Uses AI to: Analyze data to find non-obvious, mechanism-of-action based associations between compounds, molecular targets, and diseases. Allows researchers to: Reposition late preclinical stage drugs in multiple sclerosis, mitochondrial diseases, oncology, epilepsy and chronic fatigue syndrome / myalgic encephalopathy.
Uses AI to: Find applications for existing approved drugs or clinically validated candidates. Allows researchers to: Develop a pipeline of product candidates in immuno-oncology, neuroscience, and rare diseases.
Uses AI to: Match existing drugs with rare diseases. Allows researchers to: Repurpose existing drugs to accelerate treatment of rare diseases.
Uses AI to: Analyze genetic signals and molecular markers for patient response to drugs. Allows researchers to: Find clinical uses for validated cancer treatments whose development has been discontinued.
Uses AI to: Screen and reposition known drugs in unrelated indications at new, lower doses. Allows researchers to: Identify synergistic combinations of repositioned drugs for diseases with high unmet medical needs.
Uses AI to: Synthesize knowledge from multiple biomedical sources (using nference technology). Allows researchers to: Discover potential rare disease indications and subsets of patients who may respond favorably to an existing drug.
Uses AI to: Conduct experimental biology at scale by testing thousands of compounds on hundreds of cellular disease models in parallel. Allows researchers to: Rapidly identify new indications for many known drugs and shelved assets.
Uses AI to: Interpret how drug compounds would interact with people in the real world. Allows researchers to: Predict new indications for existing drugs (current focus).
5) Generate Novel Drug Candidates
Uses AI to: Curate and mine gene variants. Allows researchers to: Accelerate discovery of proteins for improving human health.
Uses AI to: Predict drug candidates by leveraging a convolutional neural network trained on a huge amount of organic chemistry data. Allows researchers to: Generate novel drug candidates faster (with a number of candidates already in development with partners).
Uses AI to: Analyze data from its "Totally Integrated Medicines Engine" platform. Allows researchers to: Perform and screen billions of assays in a single day.
Uses AI to: Ingest scientific research data sets, then form and qualify hypotheses and generate novel insights. Allows researchers to: Identify novel drug candidates (via life science-focused subsidiary BenevolentBio).
Uses AI to: Analyze data from patient samples in both healthy and diseased states to generate novel biomarkers and therapeutic targets. Allows researchers to: Generate therapeutic targets from biological data in an unbiased way, and implement personalized medicine at scale.
Uses AI to: Analyze omics data related to aging. Allows researchers to: Develop biomarkers and drugs that impact human aging.
Uses AI to: Analyze data from single-cell RNA sequencing. Allows researchers to: Understand genes in specific cells that trigger disease, then develop precision treatments for those diseases along with companion biomarker-based diagnostic tools.
Uses AI to: Search a virtual chemical space, predict binding affinity and allow filtering for drug-like properties, safety, and synthesizability. Allows researchers to: Speed drug development with a higher success rate and better targeting of hard-to-drug indications.
Cotinga Pharmaceuticals
Uses AI to: Predict biological activity from molecular structures. Allows researchers to: Intervene in pathways that cancer cells use to escape cell death.
Uses AI to: Integrate clinical trial data with real-world evidence and public datasets to eliminate silos of health information. Allows researchers to: Reduce the cost of drug development, and improve the time-to-market and likelihood of success for new drugs.
Uses AI to: Search 69 billion molecules with the goal of generating a library of 1,000 compounds to manipulate cell biology. Allows researchers to: Unlock new classes of antisense oligonucleotide therapies.
Uses AI to: Analyze RNA data from patients to identify new biomarkers and drug targets. Allows researchers to: Accelerate discovery of RNA therapeutics.
Uses AI to: Uncover gene interactions and biological networks underlying diseases, and test therapies that target them. Allows researchers to: Make analyses and predictions for precision medicine applications.
Uses AI to: Analyze complex networks of molecular interactions in cells. Allows researchers to: Acquire or in-license drug candidates.
Uses AI to: Learn best-practices from drug discovery data and experienced drug hunters. Allows researchers to: Generate drug candidates in roughly one-quarter the time of traditional approaches.
Uses AI to: Generate novel insights and predictions from biological data, chemical data, and curated databases of approved drugs. Allows researchers to: Leverage existing data to develop therapies (currently focused on infectious disease diagnostics and treatments).
Uses AI to: Simulate, filter, and search for molecules with "Generative Tensorial Networks." Allows researchers to: Discover molecules otherwise hidden from view.
Uses AI to: Design novel compounds that optimize for specific objectives. Allows researchers to: Improve the success rate of in silico to in vitro translation.
Uses AI to: Predict pharmacological properties of drugs and supplements, and identify novel biomarkers. Allows researchers to: Generate novel therapeutic candidates, with a focus on aging and age-related diseases.
Uses AI to: Shorten discovery and screening, lead optimization, and ADMET studies. Allows researchers to: Create "build-to-buy" partnerships, forming startups around new drug discovery programs that pharmaceutical companies can then acquire if successful.
Uses AI to: Analyze data in a SaaS-based bioinformatics platform for computational drug discovery. Allows researchers to: Leverage big data and machine learning for every stage of the drug discovery process, from target-identification to post-marketing activities, with no need for their own hardware infrastructure.
Uses AI to: Discover connections between drugs and diseases at a systems level by analyzing hundreds of millions of raw human, biological, pharmacological, and clinical data points. Allows researchers to: Find drug candidates and biomarkers predictive of efficacy for diseases.
Uses AI to: Analyze public and private data, with a claim to work with less data, noisier data, and more biased data than alternative approaches. Allows researchers to: Predict how a potential drug will behave in the lab and the body, with a focus on neurodegeneration, cardiovascular disease, and oncology.
Uses AI to: Predict the therapeutic potential of food-derived bioactive peptides. Allows researchers to: Cost-effectively develop highly targeted treatments for specific diseases from natural food sources.
Uses AI to: Design protein drugs through reinforcement learning. Allows researchers to: Target a wider array of binding sites, target diseases with high specificity, and create compounds that are easier to synthesize and test.
Uses AI to: Analyze many drug discovery factors simultaneously, such as effects, side effects, and toxicity. Allows researchers to: Solve complex drug discovery optimization problems.
Uses AI to: Assess and prioritize a library of drug candidates derived from analyzing tumor microenvironments. Allows researchers to: Simultaneously discover novel targets and functional antibodies for cancer.
Uses AI to: Accelerate experimentation for discovering therapies for aging and related diseases. Allows researchers to: Uncover new therapies for diseases of aging by targeting the biological processes of aging itself.
Uses AI to: Screen compound libraries for efficacy against a disease, identify new drug candidates from a public library, and identify biologic targets. Allows researchers to: Speed and reduce costs for drug discovery.
Uses AI to: Map hundreds of genes that cause a disease, then find drugs that target all at once. Allows researchers to: Find cures for complex diseases—especialy brain diseases—that involve a network of genes.
6）Validate Drug Candidates
Uses AI to: Better predict how molecules will behave in the body. Allows researchers to: Accelerate development of more effective therapies.
Uses AI to: Provide insight and analysis into a drug's poylpharmacology. Allows researchers to: Reduce attrition rates of lead therapeutic compounds and improve patient outcomes with fewer side effects.
Uses AI to: Integrate structural, target, and pathway-based descriptors to generate toxicological profiles of small molecule drugs in silico. Allows researchers to: Reduce the time and cost of preclinical decision-making while increasing the likelihood of success for compounds selected for clinical trials.
Uses AI to: Predict potency and pharmacokinetic properties of small molecules, and conceive new molecules to optimize for them. Allows researchers to: Accelerate preclinical drug development by generating and improving leads.
Uses AI to: Predict the crystalized form a drug will take. Allows researchers to: Understand the potential safety, stability, and efficacy of drug candidates.
7）Design Drugs
Uses AI to: Predict protein features and characteristics. Allows researchers to: Reduce complexity in protein design, detect production and characterisation issues, and discover novel protein features.
Uses AI to: Design peptides based on a target's solved crystal structure. Allows researchers to: Speed development of peptide drugs, which have high selectivity and low toxicity.
Uses AI to: Make and modify DNA. Allows researchers to: Prototype and edit recombinant molecules for vaccines and biologic medicines.
Uses AI to: Optimize synthetic biotherapies that are easy to manufacture, shelf stable, and outperform known antibodies. Allows researchers to: Mimic proven monoclonal antibodies with safer, more effective biologic alternatives.
8）Design Preclinical Experiments
Uses AI to: Decode open- and closed-access data on reagents such as antibodies and present published figures with actionable insights. Allows researchers to: Reduce time, money, and uncertainty in planning experiments.
Uses AI to: Determine biological variables influencing CRISPR guide design. Allows researchers to: Improve activity and reduce experimental bias in the selection of guides for CRISPR libraries.
9）Run Preclinical Experiments
Uses AI to: Automate selection, manipulation, and analysis of cells. Allows researchers to: Expedite development of cell lines and automate manufacturing of cellular therapeutics.
Uses AI to: Conduct experiments in an automated lab exactly as specified. Allows researchers to: Run experiments in a central lab from anywhere in the world.
Uses AI to: Make sense of data from testing drug candidates on their bioartificial human heart constructs. Allows researchers to: More accurately evaluate a drug candidate's cardiac safety and effectiveness.
Uses AI to: Build models to understand complex biological systems within Antha, its language and software platform for biology experiments. Allows researchers to: Optimize, reproduce, automate, and scale experiment workflows.
Uses AI to: Automate sample analysis with a robotic cloud laboratory. Allows researchers to: Generate needed data quickly and reliably with an outsourced, on-demand, automated lab.
10）Design Clinical Trials
Uses AI to: Predict which patients will respond to therapies in development. Allows researchers to: Advance therapies that fail phase 3 studies.
Uses AI to: Transform diverse streams of biomedical and healthcare data into computer models representative of individual patients. Allows researchers to: Deliver personalized medicine at scale, by revealing optimal health interventions for individual patients.
Uses AI to: Improve pathology analysis. Allows researchers to: Identify patients that would benefit from novel therapies.
Uses AI to: Automate histopathology. Allows researchers to: Stratify patients for clinical trials.
Uses AI to: Optimize clinical trial study design. Allows researchers to: Make it easier for patients to enroll and engage in clinical trials, eliminate unnecessary clinical operations burdens, and gain real-time insight into study health.
11）Recruit for Clinical Trials
Uses AI to: Make sense of unorganized and unstructured data about clinical trials. Allows researchers to: Enrol more patients in appropriate trials.
Inato
Uses AI to: Screen clinical trial sites and investigators at scale. Allows researchers to: Select appropriate sites to improve patient recruitment rates and accelerate clinical trials.
Uses AI to: Analyze medical records to find patients for clinical trials. Allows researchers to: Accelerate patient recruitment to complete clinical trials faster.
Mendel.ai
Uses AI to: Automate matching cancer patients to clinical trials through personal medical history and genetic analysis. Allows researchers to: Expedite clinical trial enrollment for cancer treatments.
12）Optimize Clinical Trials
Uses AI to: Visually confirm medication ingestion via smartphone. Allows researchers to: Improve medication adherence in clinical trials.
Athelas
Uses AI to: Analyze cancer biomarkers in 60 seconds from a drop of blood using an at-home device slightly bigger than an Amazon Echo. Allows researchers to: Optimize oncology drug development with a biomarker monitoring platform and millions of patient datapoints.
Uses AI to: Analyze structured and unstructured clinical trial participant data. Allows researchers to: Reduce clinical trial dropout rates through personalized communication.
Uses AI to: Analyze radiological images to produce clinically actionable information. Allows researchers to: Predict a patient's disease progression and treatment response, for clinical trial stratification and companion diagnostics.
Uses AI to: Find hidden health signals in data from personal devices such as laptops and smartphones. Allows researchers to: Optimize clinical trials for neurological diseases, including through better, faster identification of ideal study participants, less in-clinic observation, improved compliance, and earlier measure of drug impact.
Uses AI to: Assess and monitor cognitive health by analyzing a short speech sample. Allows researchers to: Identify patients, screen patients, and evaluate response to therapy for clinical trials of mental health treatments.
13）Publish Data
Uses AI to: Write a draft scientific manuscript based on provided data. Allows researchers to: Get a "head start" when writing a scientific manuscript to submit for publishing.
So there you have it. What did I miss? Did I get anything wrong in the descriptions? Please let me know in the comments.
作者介绍：
原文链接：https://blog.benchsci.com/startups-using-artificial-intelligence-in-drug-discovery
往期文章：
工信部2017年度中国医药工业百强榜及20研发产品线最佳工业
独角兽级公司招聘核酸/微生物/免疫/衰老/心血管和代谢疾病研究员
我的目标
1）海外生物医药人才的寻访
2）海外生物医药项目融资和引进
欢迎有兴趣的朋友和我合作，电话/微信：13564288425
备注：微信好友已上限，但是是开放接受邀请的，添加我后可以给我留言！