Treatment recommendations using genomic data and reinforcement learning

An apparatus for recommending treatment options, comprising:

• one or more memories storing processor-executable code;
• and
• one or more processors coupled with the one or more memories and individually or collectively operable to execute the code to cause the apparatus to:
• receive, for a user, data indicating a diagnosis, a genomic profile, and a treatment history;
• store, at a database with encryption at rest, a set of data comprising at least a node or vector representation that indicates the diagnosis, the genomic profile, and the treatment history for the user, wherein the node or vector representation refrains from indicating personal identifiable information (PII) for the user;
• run a rules-based engine based at least in part on the genomic profile to determine one or more treatment modifications for the user;
• input, to a reinforcement learning artificial intelligence (AI) model trained based at least in part on one or more nodes or vector representations of the set of data, one or more vectors indicating the diagnosis, the genomic profile, and the treatment history, wherein the reinforcement learning AI model outputs, in response to the one or more vectors, a plurality of treatment option recommendations and a plurality of confidence metrics, wherein a respective confidence metric corresponds to a respective treatment option recommendation;
• modify at least one treatment option recommendation of the plurality of treatment option recommendations output by the reinforcement learning AI model based at least in part on the one or more treatment modifications for the user determined by the rules-based engine;
• and
• output, to a user interface of a user device, the plurality of treatment option recommendations and the plurality of confidence metrics based at least in part on the at least one modified treatment option recommendation.

A non-transitory computer-readable medium storing code for recommending treatment options, the code comprising instructions executable by one or more processors to:

• receive, for a user, data indicating a diagnosis, a genomic profile, and a treatment history;
• store, at a database with encryption at rest, a set of data comprising at least a node or vector representation that indicates the diagnosis, the genomic profile, and the treatment history for the user, wherein the node or vector representation refrains from indicating personal identifiable information (PII) for the user;
• run a rules-based engine based at least in part on the genomic profile to determine one or more treatment modifications for the user;
• input, to a reinforcement learning artificial intelligence (AI) model trained based at least in part on one or more nodes or vector representations of the set of data, one or more vectors indicating the diagnosis, the genomic profile, and the treatment history, wherein the reinforcement learning AI model outputs, in response to the one or more vectors, a plurality of treatment option recommendations and a plurality of confidence metrics, wherein a respective confidence metric corresponds to a respective treatment option recommendation;
• modify at least one treatment option recommendation of the plurality of treatment option recommendations output by the reinforcement learning AI model based at least in part on the one or more treatment modifications for the user determined by the rules-based engine;
• and
• output, to a user interface of a user device, the plurality of treatment option recommendations and the plurality of confidence metrics based at least in part on the at least one modified treatment option recommendation.