Machine learning can be a powerful tool to recognize the symptoms caused by medical conditions. Data collected from a patient can be processed by one or more ML algorithms to determine if it matches previously collected patterns. Based upon how well the new data correlates to the patterns, the probability of a match can be calculated. The result is a list of potential diagnostic matches, along with the probability that the patient data matches previously collected patterns.

“To allow for medical diagnosis to take place at remote locations, the ML inference process was designed to run on a mobile phone device. This limited resources of a mobile device and FDA explainability requirements mandated a unique implementation process.”

To allow for medical diagnosis to take place at remote locations, the ML inference process was designed to run on a mobile phone device. This restricted the implementation space because a mobile device has fewer hardware resources and a more limited set of ML software capabilities than a large computer.