Individualizing Medicine: AI in Predicting the Effect of Yoga on Adrenomedullin Levels

I created and tuned four regression models that predicted a patient’s ADM levels after doing yoga: a Linear regressor, a Random Forest regressor, a Ridge regressor, and an Artificial Neural Network(ANN).

Non-communicable diseases(NCDs) are the biggest cause of death globally, and have no steadfast vaccine or cure. However, preventative measures such as yoga have been found to prevent NCDs as well as mediate Adrenomedullin(ADM) levels, a biomarker for NCDs, in practitioners. While this relationship has been studied, I plan to study how yoga can aid an individual’s efforts to lower their ADM levels. I did this by creating and tuning four regression models that predicted a patient’s ADM levels after doing yoga: a Linear regressor, a Random Forest regressor, a Ridge regressor, and an Artificial Neural Network(ANN). I found that the ANN had the smallest mean absolute error(MAE) and mean squared error(MSE) values, which meant that the ANN’s predicted values for ADM levels were closest to the actual values compared to the other models.

Explore More!

Research Paper