This project is inspired by 8th annual hackathon at Columbia University. Patient safety is the fundamental component in healthcare services. According to World Health Organization, the occurrence of adverse events due to unsafe care is likely one of the 10 main causes of death and disability in the world. Hartwig’s scale states that 46.03% of adverse reaction (ADRs) were classified as unpreventable, while 48.81% as probably preventable ADRs, and 5.16% as definitely preventable ADRS.

What it does

The objective of this project is to analyze the association between factors, aiming to develop an effective model to decrease possible preventable ADRs rate in practice.

How we built it

Request dataset from FDA APIs. After data cleaning and data preprocessing, we start with basic statistics analysis and then move on to logistics regression.

Challenges we ran into

The correlation between reaction count and seriousness level is relatively weak in the first place, while we determined that every drug could have one or two adverse effects, thus we just set a bottom line for the reaction count and draw relatively better conclusions.

Accomplishments that we're proud of

We use model to confirm finding coming from data analysis, and deliver insight to the industry.

What we learned

How to brainstorm and handle complicate dataset in a short time. How to teamwork to complete the challenging project.

What's next for Influential Factors of Patients’ Serious Adverse Effects

There are plenty of data in the original dataset we can further utilize The connection between various adverse effects and drugs can be explored if we have time to construct a proper mapping/connection relationship between these two factors We can try more powerful models in the future

Built With

Share this project: