Identifying Existing Non-Cancer Generic Drugs Through AI

Background

Cancer remains a global health crisis, with 17 million new diagnoses, 10 million deaths, and $1 trillion in costs annually. Despite these challenges, many non-cancer generic drugs have shown promise in cancer treatment. However, identifying these opportunities requires analyzing unstructured clinical study data. Reboot Rx, a tech nonprofit, sought to streamline this process using AI to repurpose affordable generic drugs for cancer care.

Objective

The project aimed to develop automated methods to extract, classify, and structure key information—such as patient counts and study outcomes—from clinical study abstracts. This structured data would identify promising generic drugs for repurposing in cancer treatment.

Approach

The team of 50 technology changemakers collaborated with Reboot Rx to address the challenge. Key steps included:

• Data Sources: Leveraging study abstracts provided by Reboot Rx, which included study IDs, titles, and abstracts.
• Analysis Techniques: Implementing supervised and semi-supervised modeling techniques combined with rule-based methods to enhance data extraction accuracy.
• Tools and Methods: Creating NLP pipelines to extract numerical values (e.g., response rates, durations) and label them with corresponding outcome measures. Curated training datasets helped improve model performance.
• Scope Management: Focusing on text from study abstracts and predefined outcome measures (e.g., percentages, months).

Results and Impact

The project successfully developed an automated NLP pipeline to extract and classify data from clinical study abstracts. Key outcomes included:

• Creation of a scalable, structured database containing outcome measures.
• Machine learning models capable of classifying defined variables with high accuracy.
• A curated dataset of annotations for further research and analysis.

This structured approach accelerates the identification of non-cancer generic drugs that can be repurposed for cancer treatment, reducing costs and increasing accessibility to life-saving therapies.

Future Implications

This project demonstrates the power of AI in transforming unstructured clinical data into actionable insights. The findings could influence future cancer treatment policies, streamline drug repurposing efforts, and inspire further research in automating complex data extraction tasks. Reboot Rx’s innovative approach sets a precedent for leveraging technology to address global health challenges.