Prediction story

Prediction story: Is it beneficial to transition from business rules to Machine Learning (ML)

Predicting patient lines of therapy is crucial, particularly in fields like Oncology, Neurology, and rare diseases. Biopharma companies can leverage this information to proactively engage prescribers when patients become eligible for specific treatments.

In 2018-19, my company undertook a project to predict patient transitioning from Line of Therapy (LoT) 1 to LoT 2 using business rules. These rules considered factors such as the medication taken, therapy duration, progression-free survival (PFS), and prescribing patterns of healthcare providers (HCPs). 

While it was effective in the past, the rise of ML has made predictions based on business rules less favorable due to inherent challenges in it like:

> Inflexibility: Difficulty adapting to evolving patient data.

> Lack of granularity: Insufficient detail to capture the subtle nuances of individual patient profiles, resulting in less personalized predictions.

Machine Learning (Random Forest algorithm) offers enhanced accuracy by learning from extensive patient data, including medical records, genetic information, and imaging results, enabling earlier and more precise predictions.

What has been your experience with using machine learning for patient prediction?

"Follow" me for more such stories

Read past stories like above at https://lnkd.in/d9ijze8r