Inspiration

My summer internship at a biotech company inspired me to look out for ways to make the exclusive drugs more affordable to the patients as well as for the manufacturers. Biotechnology is a precarious field where there is a tension between monetary funding for novel scientific innovation and affordability of the drugs. We tried to break down the lifecycle of a drug from start to end and found an immense potential to optimize the clinical trials process.

What it does

The application intends to leverage the clinical trials database and Large Language Models to help manufacturers make faster decisions to pick potential clinical trial sites (by considering the population demographics and the feasibility) and also inform the manufacturers about potential adverse effects shown by the participants. This could save $50k for each clinical trial phase (for each drug!) and choose diverse participants, thereby improving the efficiency of the clinical trials.

How we built it

We built a chat interface wherein the manufacturer representative can ask for questions by providing the context of their clinical trial that they intend to conduct. Zero-shot learning in Large Language Model is used along with the publicly available clinical trials dataset to identify potential adverse effects that'd be perceived by the participants.

Challenges we ran into

We wanted to personalize the choice of site location based on the demographic mixture. Given the tight deadlines, we couldn't get access to participant wise clinical trial information. Combining this data along with the exisiting clinical trial output would unlock more opportunities to save cost.

Accomplishments that we're proud of

The app accurately identifies the potential side effects for most of the medical conditions with zero-shot learning ( without having no context).

What we learned

We learned about the complex world of clinical trials and drug manufacturing process. This being a regulated space, it pushed our team to think of innovative ways to overcome this barrier.

What's next for Clinical Trials Expert

Getting access to participant-wise clinical trials data and improving the capabilities of the LLM specific to Clinical Trials and Medical Drugs.

Built With

Share this project:

Updates