Using AI to improve efficiency and effectiveness during clinical trials
An overview of AI/ML tools, startups and companies transforming drug development
“Positive findings are around twice as likely to be published as negative findings. This is a cancer at the core of evidence-based medicine.”
Ben Goldacre #AllTrials
A 2019 report from Deloitte found that the average cost of bringing a drug to the market increased from $1.188 billion in 2010 to $1.981 billion in 2019, while the average forecast peak sales per asset declined from $816 million in 2010 to a low of $376 million in 2019. As a result, the expected return on investment decreased from 10.1% in 2010 to 1.8% in 2019. These declining returns, due to productivity challenges across the discovery and clinical development phases of drug development, call into question big pharma’s R&D model.
Therefore, due to the excessive cost, the delays, the failures of clinical trials and the declining returns on investment, more effective and efficient ways of conducting clinical trials are required urgently. On account of this, AI, deep learning (DL), machine learning (ML) and natural language processing (NLP) combined with an appropriate digital infrastructure, they all have now the potential to improve clinical trial rates and reduce development costs. Accordingly, the AI-Based Clinical Trial Solution market which was $1.3 billion in 2022, would rocket up to $5.55 billion by 2030.
So, today’s newsletter is an overview of the different AI/ML tools, startups and companies transforming the traditional drug development process, from clinical trial research and data mining, to pharmacoepidemiology and pharmacovigilance.
❇️ Clinical trial research
1️⃣ AI Protocol Design and Reporting:
💡 The SPIRIT-AI extension was developed for clinical trial protocols in parallel with its companion statement for trial reports 💡 CONSORT-AI (Consolidated Standards of Reporting Trials-Artificial Intelligence). SPIRIT-AI can assist editors and peer reviewers, as well as the general readership, to understand, interpret, and critically appraise the design and risk of bias for a planned clinical trial.
💡 Concerto HealthAI, uses its RWD capabilities and eurekaHealth AI platform to design robust clinical trials and generate precision treatment insights.
💡 Citeline’s Trialtrove supports trial decision-makers throughout the clinical trial life cycle, from strategy and design to execution. Drawing from over 60,000 sources, Trialtrove provides unmatched trial intelligence, curating data on trial benchmarks and metrics, enrolment and study timelines, patient populations, endpoints, outcomes, geographic distribution and many more.
💡 Trials.AI uses AI to analyse large sets of genomic data, journal articles, past clinical studies and other forms of research to improve study design. The system is able to unlock information, derive insights and make recommendations to trial sponsors on how to best design and optimise their trial protocols, as well-designed protocol limits improves recruitment and retention and reduces burden on patients and trial sites by bringing in cost and time efficiency.
2️⃣ AI Patient Selection and Recruitment:
AI solutions for identification, matching and patient recruitment:
💡 IBM Watson for Clinical Trial Matching (CTM) can collect and link structured and unstructured data from EHRs, medical literature, trial information and eligibility criteria from public databases like ClinicalTrials.gov. For example, early use of Watson for CTM by Mayo Clinic resulted in 80% increase in enrolment in clinical trials for breast cancer. Moreover, Mayo and IBM Watson Health plan to develop this system further for other applications, such as radiation, surgery and supportive care.
💡 Antidote platform uses ML to seamlessly connect patients with clinical trials, creating a simple process for both drug sponsors and the patients they need to reach. Their system uses ML to structure and organise trial listings from ClinicalTrials.gov into a patient-friendly, searchable format.
💡 MENDEL.AI uses AI tools to analyse clinical data, including medical history and genetic analysis from cancer patients. Its goal is to facilitate clinical trials in oncology research by organising and analysing RWD, and matching patients to the right clinical trials.
💡 ResearchMatch is a free secure registry, making it easier for the public to volunteer and to become involved in clinical research studies. The tool is expected to reduce recruitment costs, increase enrolment and speed research progress.
💡 Trial Pathfinder is an AI framework to evaluate clinical trial eligibility criteria.
💡 Deep6AI is the leader in precision matching research software, connecting all research stakeholders in an AI-powered real-time ecosystem. Deep6AI mines millions of patient records, including “unstructured” physician notes, to pinpoint eligible patients in real time, and has just announced the launch of its genomics module that uses AI to mine genomics data across an entire EMR system and disparate genetic reports to find patients with specific genetic markers in real time.
💡 Criteria2Query (C2Q) is an automatic cohort identification system that enhances human-computer collaboration to convert complex eligibility criteria text into more accurate and feasible cohort SQL queries.
💡 DQueST contributes a novel framework for transforming free-text eligibility criteria to questions and filtering out clinical trials based on user answers to questions. In particular, DQueST transforms criteria narratives in the ClinicalTrials.gov repository into a structured format, normalises clinical entities using standard concepts, clusters related criteria, and stores the resulting curated library.
Medical coding is a critical process that involves assigning standardised codes to medical terms in clinical trial participant medical records, diagnoses and procedures, in order to accurately track and record a patient’s treatment data and that the research study is conducted ethically and efficiently. Some Medical Coding Generative AI Startups are:
💡 Fathom Health is a leader in autonomous medical coding, applying cutting-edge deep learning and NLP to code patient encounters with the highest automation rates and the broadest specialty coverage.
💡 BUDDI.AI is the most accurate medical coding automation platform on the market, offering CODING.AI.
💡 CodaMetrix is a multi-specialty coding AI-platform that translates clinical information into accurate sets of medical codes for patient care and revenue cycle processes, from fee-for-service to value-based care models.
💡 Maverick’s platform mCoder™ offers a major improvement over standard rule-based and pattern recognition solutions currently available on the market.
💡 Arintra is an holistic solution for fully automating the coding workflow and streamlining the revenue cycle.
💡 Nym is solving the challenges associated with medical coding by fully automating the coding process.
3️⃣ AI Investigator and Site Selection
💡 Telisina, a data strategy consultancy company, utilised AI and ML to drive site identification and activation in clinical trials for a rare genetic disease with: FormaAI (that identified precise patient characteristics within the rare disease population), IndiciumAI (that ranked trial sites and investigators by capability to function effectively), PopEstimateAI (a predictive model to calculate the patient population amidst COVID-19) and CognitusAI (that mapped patients to viable key investigator and trial sites).
4️⃣ AI Monitoring and Management of Clinical Trials
AI algorithms analyse real-time patient data, ensuring trial integrity and identifying adverse events.
In 2016, GSK launched the 💡PARADE (Patient Rheumatoid Arthritis Data from the Real World) study for which it developed an app using Apple’s ResearchKit framework.
In 2018, Novartis announced the launching of 💡FocalView, an app developed through Apple’s ResearchKit to be used as an ophthalmic digital research platform by allowing clinical researchers to monitor disease progression.
💡 AiCure is a clinical trial management platform offering: AiCure Patient Connect (a mobile app used by participants for real time dosing instructions and leveraging computer vision to ensure medication adherence), AiCure’s Platform (a regulatory-compliant AI-platform that captures and analyses digital biomarkers for better understanding and prediction of participant behaviour and response to treatment) and AiCure Site Services (to support clinical operations and patient engagement).
Generative AI startups for monitoring:
💡 Tytocare is offering TytoPro, an FDA-cleared device and adapters (otoscope, stethoscope 🩺 and tongue depressor adaptors) for remote physical exams. Tytocare just announced that it has raised $49 million in additional growth funding and that has signed a commercial collaboration with Light Vortex Division, the digital activities arm of Japanese insurance giant Sompo, to focus on nursing homes and the elderly care market in Japan
💡 Athelas’s Home is an at-home FDA-cleared blood diagnostics device that helps simplify and enhance the blood monitoring process. In 2022, Athelas hit unicorn status after raising $132 M in two consecutive funding rounds that boosted the company's valuation to $1.5 billion.
💡 BioFourmis’s platform offers remote data collection, digital endpoints from Biofourmis' library of biomarkers, digital tools support, trial decentralisation, safety monitoring, customised care, and easier and equitable patient recruitment options.
💡 Current Health provides an enterprise care-at-home platform, enabling a clear window into patient health at home and responsive remote care.
Phlexglobal, a PharmaLex company, is the recognised a global authority in Trial Master File 🗄️ (TMF) services (a compilation of documents that prove that the clinical trial has been conducted following regulatory requirements including Good Clinical Practice). Hundreds of sponsor and CRO companies worldwide rely on Phlexglobal’s next-generation eTMF solution and more than 150 experts to elevate TMF management, reduce inspection risk and solve TMF challenges.
5️⃣ AI Image Analysis and Biomarker Computation
💡 Paige, a global leader in end-to-end digital pathology solutions and clinical AI with the first Large Foundation Model using over one billion images from half a million pathology slides across multiple cancer types, is developing with Microsoft a new AI model that is orders-of-magnitude larger than any other image-based AI model existing today configured with billions of parameters. For this collaboration, Paige is incorporating up to four million digitised microscopy slides across multiple types of cancer from its unmatched petabyte-scale archive of clinical data, and will utilise Microsoft’s advanced supercomputing infrastructure to train the technology at scale and ultimately deploy it to hospitals and laboratories across the globe using Azure.
💡 Altis Labs is a computational imaging company accelerating clinical trials with its AI platform Nota, and enables sponsors to ingest imaging data from sites and CROs, creating the single source of truth for imaging data. This summer Altis Lab announced the closing of its $6 million seed financing.
💡 The Biognosys TrueSignature™ platform provides high-precision customisable proteomics panels for pharmacodynamic readouts and clinical biomarker monitoring. Last year, Biognosys, announced TrueDiscovery platform offering insights on 4,200 proteins in plasma, 11,000 proteins in other biofluids, and 13,800 proteins in tissue, covering all relevant disease pathways.
6️⃣ AI Intelligent Data Collection and Management
In 2016, Takeda established the R&D Data Science Institute, that integrates data sets such as clinical trials, observational studies, population-level biobanks and RWD. Then Takeda used Deloitte’s 💡 ConvergeHEALTH Deep Miner™ platform to analyse massive amounts of RWD. Moreover, the AWS Cloud-enabled Deep MinerTM platform used DL and ML techniques to improve the predictive accuracy of Takeda’s earlier results from 53.4% to 92%.
💡 Glass Health, a digital notebook for clinicians, just raised a $5M round led by Initialised Capital. They use a technique called retrieval augmented generation to connect a large language model to their database of clinical guidelines created and maintained by the Glass Health Clinical Team of academic physicians.
💡 Medidata is offering the Medidata Clinical Cloud a leading unified platform dedicated to clinical research and the Data Fabric a powerful, cutting-edge architecture behind the Medidata Clinical Cloud. They offer protocol development, clinical site collaboration and management; randomisation and trial supply management; capturing patient data through web forms, mobile health (mHealth) devices, laboratory reports, and imaging systems; quality monitor management; safety event capture; and monitoring and business analytics. So far, more than 30,000 studies have been run on the cloud computing platform from Medidata Solutions.
Other Generative AI startups for notetaking in clinics are 💡: eleos health, DeepScribe, Suki, Health Note, Ambience, Mentalyc and Knowtex.
❇️ Data Mining: aggregation and synthesis of biomedical data, generating models and analysing real world evidence during clinical trials
💡 Cascade MD enables healthcare providers to dictate and capture the entire patient visit details with their mobile devices, by using Cascade MD's voice-to-text and AI inferencing engine to capture all important information in real-time and integrate it into an EMR.
💡 Unlearn develops generative ML methods to predict individual health outcomes and accelerate clinical innovation, by using AI-powered digital twins on every patient. Just this summer QurAlis, a biotech company focused on developing precision medicines for amyotrophic lateral sclerosis, announced that it had partnered with Unlearn to accelerate and optimise clinical program by using Unlearn’s advanced ML algorithms to create digital twins or synthetic versions of patients enrolled in clinical trials.
💡 GNS Healthcare (now AITIA BIO) uses its Reverse Engineering & Forward Simulation (REFS) or Gemini Digital Twins namely a causal AI and simulation platform, to analyse RWD and clinical trial data to model patient responses to treatments in silico and has worked with biopharma companies, including Amgen, BMS, Celgene, Johnson & Johnson and Novartis.
💡 Datavant, which specialises in breaking down silos 🧱 and analysing health data securely and privately, acquired Swellbox to enable patients to request their medical records seamlessly. Swellbox also enables patient authorisation for record retrieval for clinical trial recruitment, long-term surveillance, registry creation and other use cases. Datavant is now working with Amazon Web Services (AWS) to manage and optimise clinical and patient insights.
💡 Linguamatics (an IQVIA company, the advanced analytics, technology solutions and clinical research services to the life sciences industry) is the market leading NLP processing AI platform for Healthcare & Pharma offering among other things text mining, also referred to as text analytics. Namely, an AI technology that uses NLP to transform the free unstructured text in documents and databases into normalised, structured data suitable for analysis.
💡Syapse is a data intelligence company powered by Syapse Raydar, that leverages the power of NLP and ML to transform highly contextual, clinically complex oncology data into actionable insights. Last year, Syapse entered in collaboration with Merck to leverage Syapse’s leading real-world evidence solutions designed to improve outcomes for cancer patients.
💡AETION just announced its 'Screening Tool to Evaluate whether Using Real-world Data to Support an Effectiveness Claim in an FDA Application has Regulatory Feasibility’ (or the SURF tool), in order to help sponsors make an initial feasibility assessment for using RWD to provide substantial evidence of effectiveness to support FDA approval.
💡Saama automates key clinical development and commercialisation processes with AI, ML and advanced-analytics accelerating time to market. This summer Saama announced the launch of its unified platform of SaaS-based products (to accompany its existing portfolio of customised solutions and services), that reduces query identification and generation times by as much as 90% per query, realises up to 50% time savings for study data transformations, reduces time from data entry to analysis by more than 35%, and many more.
📌 Pharmacoepidemiology and pharmacovigilance
1️⃣ AI Models for Predicting Drug Response and Optimising Treatment Outcomes
By applying a DL-based approach using Bidirectional Encoder Representations from Transformers (BERT) models on a dataset of 10,000 reviews from WebMD and Drugs.com, 💡 a proposed model achieved state-of-the-art performance in adverse drug events (ADEs) detection and extraction.
2️⃣ Integration of Genomic Data and AI Algorithms for Personalised Drug Selection
💡 MEDIDATA’S ACORN AI is focusing on addressing two big trends: the first is precision medicine with CAR-T therapy, tissue engineering, gene and cell therapy, and the second is the advancement of AI and new sources of data to give researchers a 360-degree view of the patient, including clinical, genomic, molecular, as well as socio-economic, behavioural and environmental data. The company just announced the launch of its Medidata diversity program, aimed to improve diversity, equity and inclusion in clinical trials, by addressing common systemic and clinical barriers that can hinder a diverse range of individuals from participating in clinical trials.
💡 Genoox is offering solutions for Clinical Genetics combining public data with community data to streamline the path from DNA sample to clinical report, by using Genoox’s cloud-based AI platform Franklin.
3️⃣ AI-Guided Dosage Optimisation for Individual Patients
💡 CURATE.AI is an AI-powered platform for personalised medicine that optimises treatment outcomes by considering individual patient characteristics. The platform dynamically adjusts chemotherapy doses for cancer patients, aiming for optimal efficacy and minimal toxicity. CURATE.AI has potential applications in hypertension, diabetes, and digital therapeutics.
💡 Tempus is an AI-enabled precision medicine that combines DNA sequencing for cancer with AI. FDA has granted Tempus Breakthrough Device Designation for its HLA-LOH assay as a companion diagnostic (CDx) test, that uses ML to analyse sequence data produced by Tempus’ FDA-approved next generation sequencing-based xT CDx assay, in order to identify cancer patients with solid tumours who may benefit from treatment with specific targeted therapies.
Finally, just this week the AI and analytics software company SAS announced its collaboration with AstraZeneca to increase efficiency and drive automation in the delivery of statistical analyses for clinical and post-approval submissions to regulatory authorities.
Until next time,
📄 References: