System, Method and Apparatus for Forming Machine Learning Sessions

A system to dynamically classify toileting acts of a user using a temporal machine learning architecture, the system comprising:

• a processor circuitry;
• and
• a memory circuitry in communication with the processor circuitry, the memory circuitry configured with instructions to cause the processor circuitry to:
• receive a plurality of frames from one or more sensors capturing the user's toileting activity, each frame corresponding to a discrete time interval;
• form one or more sequences by grouping the frames separated from each other by a substantially constant or dynamically adjustable time differential;
• extract, from each frame, one or more feature vectors including optical, capacitive, depth-field, or impedance parameters corresponding to the toileting act;
• input the feature vectors into a temporal machine learning model configured to identify one or more conditions associated with the toileting act, wherein the temporal machine learning model comprises at least one neural network selected from a recurrent neural network, a long short-term memory (LSTM) network, or a transformer network;
• assign to each identified condition a confidence score indicative of a likelihood that the condition is present;
• generate a session-level annotation record comprising the one or more conditions, the confidence scores, and feature attribution data identifying features contributing to the classification outcome;
• and
• communicate the session-level annotation record to a remote repository when the confidence score or a condition threshold exceeds a predetermined value.