Forty Winks

Forty Winks estimates the probability that a user's sleep patterns are correlated with one of the three most common sleep disorders: insomnia, narcolepsy, or sleep apnea. Data provided from the user's fitness tracker is fed into a Bayesian network, a probabalistic model that uses inference to calculate the probability of a disorder given a network of symptoms. The symptoms and causal relationships are weighted based on available diagnostic criteria for medical professionals and publicly available statistics.

Users have the opportunity to provide feedback on the accuracy of the predictions. The feedback can be used to train an artificial neural network, which applies machine learning techniques to increase diagnostic accuracy and allow doctors to better understand the weights and relationships of symptoms. Over time, Forty Winks will become more accurate and have the potential to provide statistical data to doctors and researchers to improve healthcare.

Created for Hack Arizona 2016.