AI-Powered Clinical Trial Matching for BIPOC Health
Try it out: https://trialmatch-ai-0451d1f7fd04.herokuapp.com/
Inspiration 🌟
Did you know that even though diseases like diabetes, hypertension, and certain cancers disproportionately affect communities of color, participation in clinical trials by these groups is shockingly low? 😮
Here's some eye-opening data:
Breast Cancer:
Incidence: African American women have the second-highest incidence rate at 126.7 per 100,000, but the highest mortality rate. 🎗️
Clinical Trial Participation: Only 6% of participants are African American women, while White women make up 80%. This underrepresentation affects treatment effectiveness for BIPOC women. 😔
Prostate Cancer:
Incidence: African American men have the highest incidence rate at 178.3 per 100,000, more than double that of Asian/Pacific Islander men at 55.3 per 100,000. ⚕️
Clinical Trial Participation: Despite higher incidence, African American men represent only 9% of trial participants, whereas White men account for 81%.
This lack of representation means that treatments may not be as effective for these communities, perpetuating health disparities. 😔
We wanted to change this! 💪 Our mission is to empower BIPOC (Black, Indigenous, and People of Color) communities by connecting them to clinical trials through the power of AI.
What It Does 🚀
TrialMatch.ai is a user-friendly web app that provides interactive data visualizations and helps patients from BIPOC communities find clinical trials they might be eligible for! 🩺✨
- Simple Input: Users enter their age, gender, race/ethnicity, location, disease, and any additional info.
- AI-Powered Matching: Our app uses AI to read through complex clinical trial eligibility criteria and determine if the patient is a good match. 🧠
- Quick Results: Eligible trials are presented with easy-to-understand explanations, making it simple for users to see why they are a match.
How We Built It 🛠️
We combined several awesome technologies to make TrialMatch.ai a reality:
- Python & Flask: For building the web application.
- OpenAI GPT-4: To intelligently match patients with clinical trials by understanding eligibility criteria. 🤖
- Celery & Redis: To handle background tasks efficiently, ensuring the app responds quickly without timeouts. ⏳
- ClinicalTrials.gov API: To fetch up-to-date clinical trial data.
- Data Visualization: We used HTML, CSS, JavaScript, and Chart.js to create interactive charts that highlight health disparities in underserved communities. 📊
Here's a snippet of our main code:
DISEASES_OF_INTEREST = [
"Sickle Cell Disease", "Type 2 Diabetes", "Hypertension", "Breast Cancer",
"Prostate Cancer", "Colorectal Cancer", "Asthma", "Obesity", "HIV/AIDS"
]
def get_clinical_trials(search_terms):
base_url = "https://clinicaltrials.gov/api/query/full_studies"
params = {
'expr': search_terms,
'min_rnk': 1,
'max_rnk': 10,
'fmt': 'json'
}
Challenges We Ran Into 🏃♀️
- Complex Eligibility Criteria: Clinical trial requirements are detailed and sometimes hard to interpret. Teaching the AI to understand and evaluate them accurately was challenging.
- Handling Timeouts: Initial requests were taking too long, causing timeouts on Heroku. We implemented Celery to process tasks in the background, ensuring users didn't have to wait long. ⏰
- Data Visualization: Presenting complex health data in an engaging and understandable way required careful planning and design.
Accomplishments That We're Proud Of 🏅
- Empowering BIPOC Communities: Creating a tool that bridges healthcare gaps for BIPOC community. ❤️
- AI Integration: Successfully using AI to simplify complex medical data and match patients with clinical trials.
- Interactive Charts: Our visualizations effectively highlight health disparities, raising awareness and educating users. 📈
What We Learned 📚
- Health Disparities Are Significant: We gained a deeper understanding of the health challenges faced by underserved communities and the importance of representation in clinical trials.
- Technical Skills: Learnt a ton of Flask, Celery, Redis, and integrating OpenAI models into applications. 🚀
- User-Centered Design: Learned the importance of making technology accessible and easy to use, especially for users dealing with complex health information.
What's Next for TrialMatch.ai 🌈
- Use HIPAA-compliant LLMs: Use AI models that are compliant with healthcare laws to protect confidentiality
- Expand Diseases Covered: Include more conditions to help a broader audience.
- Improve UI/UX: Make the app even more intuitive and visually appealing. 🎨
- Add Multilingual Support: Implement support for other languages to reach non-English speakers. 🌍
- Community Partnerships: Collaborate with healthcare providers to maximize impact and reach more people. 🤝
- Mobile App Development: Create a mobile version of the app for on-the-go access.
We're excited about the future of TrialMatch.ai and its potential to make a real difference in reducing health disparities! Thank you for being part of my journey. 😊
Note: All data visualizations and statistics are sourced from trusted organizations like the CDC, NIH, American Heart Association, and more (2021-2023 data).
Let's bridge the gap and empower BIPOC communities with better access to clinical trials! 🌉❤️



Log in or sign up for Devpost to join the conversation.