Systems and methods for an artificial intelligence health coach with biometric data integration and personalized health intervention generation

A non-transitory computer-readable medium storing instructions that, when executed by one or more processors of a device, cause the device to:

• receive a user input;
• interpret the user input with natural language processing performed by an artificial intelligence model that includes a large language model to determine a meaning of the user input, including whether the user input requests personalized health intervention information;
• receive a glucose level from a continuous glucose monitor;
• receive an insulin level retrieved from a database;
• calculate a metabolic health index from the glucose level and the insulin level, the metabolic health index including at least an insulin resistance index calculated from the glucose level and the insulin level;
• analyze the glucose level, the insulin level, and the metabolic health index with the artificial intelligence model;
• identify, based on the metabolic health index and the meaning of the user input, a physiological trend of the user;
• compare the metabolic health index to historical metabolic health index data associated with the user to determine whether the metabolic health index has changed over time;
• generate, using the artificial intelligence model, personalized health intervention information relating to the identified physiological trend of the user and modified based on whether the metabolic health index has changed over time;
• and
• output the personalized health intervention information to the user.

A health coach system comprising:

• a voice interface configured to receive a user input including at least one of a spoken input through a microphone, a typed input through a keyboard, or a gestural input through a camera, the voice interface including computer vision models configured to (a) recognize and interpret gestures and facial expressions, and (b) to process the user input using natural language processing, wherein when the user input includes the spoken input the voice interface performs prosody analysis on the spoken input to detect emotional cues;
• one or more sensors including a continuous glucose monitor configured to provide a glucose level to the health coach system;
• one or more data sources configured to provide an insulin level retrieved from a database;
• one or more processors configured to calculate a Homeostatic Model Assessment of Insulin Resistance score from the glucose level and the insulin level, the one or more processors further being configured to compare the Homeostatic Model Assessment of Insulin Resistance score to historical Homeostatic Model Assessment of Insulin Resistance scores associated with the user to determine whether the score has changed over time;
• a large language model configured to receive the processed user input, the biometric data, and the calculated Homeostatic Model Assessment of Insulin Resistance score, the large language model being further configured to analyze the biometric data and identify a physiological trend or condition of the user, the large language model being further configured to generate personalized health intervention information relating to the identified physiological trend or condition and modified based on whether the Homeostatic Model Assessment of Insulin Resistance score has changed over time;
• a user data storage configured to store conversation records, recorded emotional states, biometric data history, and user preferences as a persistent user profile;
• and
• an output interface configured to output the personalized health intervention information to the user through at least one of a speaker or a screen;
• wherein the health coach system uses the persistent user profile as contextual memory to interpret user inputs and generate personalized health intervention information across multiple interaction sessions.