Maximizing Clinical Trial Access With AI

Submitted by Jack Owens | Icarus Therapeutics.
Originally published in Life Sciences Insights Magazine, August 2025

Clinical trials make a salient barrier to the drug approval process for pharma and biotech companies. Upwards of 70% of investor funding at pharma and biotech startups is allocated toward the clinical approval process, and it takes an average of 10–15 years and $1–2B to bring a drug to market. Within the clinical pipeline, a significant bottleneck is patient recruitment and retention. Despite the fact that suitable trials exist for 85% of patients whose most viable treatment option would be a clinical study, nine in ten trials still fail to meet their enrollment timelines, and more than a quarter of research sites fail to enroll even a single participant. Meanwhile, patients are faced with an arduous process to find a clinical trial that both targets their specific condition and has eligibility requirements that they satisfy, all the while faced with an impending deterioration in their health.

The primary database patients have to search for clinical trials, ClinicalTrials.gov, does not offer an easy way for patients to locate trials based on their specific background. Searches on ClinicalTrials.gov are often based on keywords that are arbitrarily chosen by researchers who enter trial information into the website, making it easy to overlook relevant studies. Search results are not organized based on any apparent relevance, making it difficult to decide on a trial without further investigation. Furthermore, the search feature provides minimal means to rule out trials for which the patient would be ineligible. Sex and age are the only inclusion and exclusion criteria a user can easily tailor searches to, and other eligibility criteria, much like with keywords, are not standardized across studies. Ultimately, users have to carefully review eligibility criteria line by line in order to determine whether a study is relevant or not. Patients who consult with doctors to search ClinicalTrials.gov for them rarely receive consistent recommendations, often getting completely different sets of trials from each.

When patients instead seek direct referrals to clinical studies, most doctors do not have specific trials to recommend, and tend to refer patients to large institutions with strong reputations that aren’t necessarily running the most suitable trials. The only truly effective option to find a trial is by hiring an expert to spend hours searching for and reviewing studies by evaluating the results of preclinical and previous study phases, perusing drug companies’ financial information that may suggest positive unpublished findings, identifying each study’s specificity to the patient’s particular ailment, and comparing trials to successful past research — a very costly process that still is not guaranteed to identify the most optimal studies.

Icarus Therapeutics is a TechBio startup developing an end-to-end platform for both the provider and site facing sides of clinical recruitment that leverages large language models (LLMs) to rapidly digitize patient records and clinical documentation with a core matching engine that uses said data to automatically pair patients with trials and also provide detailed reasoning behind each referral, allowing for a user-friendly review and final decision on every match. Through the use of natural language processing (NLP), what had previously required days to weeks of careful review by multiple research site employees can be done in a fraction of the time without any direct human involvement. 80% of patient data is found in doctor’s notes, which have minimal structure and consistency across a patient’s journey. The unstructured and text-heavy nature of this data makes it ripe for analysis with NLP. In addition to patient data, LLMs can also expedite the analysis of trial documentation. Clinical trial protocols outline all aspects of a study and come in the form of large text documents. These protocols can be up to hundreds of pages long, which makes evaluating trials by humans a grueling process on both sides of recruitment. LLMs also come into play within the matching engine, used for cross checking trials’ extensive eligibility criteria against thousands of patients quickly and for translating the internal logic behind each match into a human-readable format, helping to get around AI’s black-box problem.

A common issue facing artificial intelligence and machine learning is that when a model is trained to recognize patterns in data, the internal rules it produces to accurately analyze inputs are often difficult to identify or explain. This lack of transparency raises many practical and ethical concerns, especially within a field as serious as healthcare. To address this problem, the Icarus Therapeutics’ software provides a detailed synopsis behind every match it makes such that a human reviewer at a research site can always verify whether a particular patient is well-suited to their study, and a patient will have better access to information when consulting with their doctor to choose whether a particular trial is right for them.

While not a complete replacement for human decision making in selecting patients or trials, Icarus is a powerful tool that provides users with better information that will allow them to make more informed decisions within a shortened time frame. On the provider side, users can obtain a brief list of studies organized by relevance, each with a concise explanation as to why it’s applicable, sparing them the taxing effort of sifting through thousands of clinical protocols and their supplementary information. On the site side, users can receive a highlighted description of every prospective patient’s history as it relates to each trial, avoiding hours of meticulous review of patient medical records. With future prospects to incorporate more complex patient information such as genetic signatures and genealogy records and to provide insights beyond recruitment to allow real-time optimization of studies at sites, the working Icarus Therapeutics platform demonstrates AI’s profound capacity to accelerate the drug discovery process through a highly user-friendly, end-to-end matchmaking tool.

Get the latest thought leadership from California’s life sciences sector in the quarterly Life Sciences Insights magazine. Share your own news and insights with CLS Member Voice.