Patient Similarity Networks Development to Guide Clinical Decision-Making

The patient similarity is a way to use medical data sets to integrate multiple data types to facilitate medical decision-making. While individualized data-driven clinical analysis is still developing, patient similarity networks can facilitate research. Moreover, patient similarity solutions are crucial to innovative targeted drug trials, overcoming drug resistance, and combining therapies. Patient similarity networks development is crucial in the case of hydroxychloroquine, as it has got negative coverage due to adverse events when treating patients with COVID-19 pneumonia for emergency use. Our client wanted us to define whether hydroxychloroquine is safe for patients with rheumatoid arthritis (at the age of 18 years or older).

• First of all, we had to convert the source data into a standard machine-readable format by performing a full ETL cycle and carry out initial statistical analysis to make a propensity score estimation.
• Then we choose the matching algorithm: сlustering, dimensionality reduction, and similarity.
• After all, we assessed matching quality, treatment effects, and their standard error estimation.
• Finally, a sensitivity analysis of estimated treatment effects with respect to unobserved heterogeneity was done.

As a result, we defined that hydroxychloroquine treatment is safe for patients with rheumatoid arthritis in the short term but might be associated with cardiovascular mortality in the long term.

1. Domain Expertise in Healthcare
2. Natural Language Processing (NLP)
3. Relational Database Management (SQL-based)
4. OHDSI Software Tools (Athena, Whiterabbit and Rabbit-in-a-hat, Usagi, Data Quality Dashboard, Achilles, Atlas)
5. Data Science and Data Analytics
6. Programming languages: R, Python, SQL
7. Machine Learning
8. Predictive Analysis