HEART: Adding Memory to Reasoning: macHine lEArning tailoRed drug Titration

HypertensionHTNaffectsmillion Americanswith an annual hospitalization cost of $billion yearClinical trial outcomes of optima BPO BPa clinical reasoning artificial intelligenceAIsystemdetermined that the success failure of past anti HTN medication treatments can be utilized to improve O BP predictive performanceOur objective for this Phase I proposal is to investigate the feasibility of building a past treatment success failurememoryfunction into O BPO BPan innovation of Optima Integrated Healthis functional at UC San Francisco Medical CenterUCSF MCthe clinical collaborating partnerO BP AI safely optimizes HTN medication treatment independent of in office visitsIt collects and analyzes current patient health information from multiple sources to determine if a medication optimization is requiredO BP then alerts the physician of a need for a medication change by providing a clinically supported treatment recommendation actionIn this Phase Iwe propose the development and validation of O BPandapos smemoryby using the principles of Instance Based LearningIBLKnown for its flexibility and intuitive logicthe IBL theory will be adapted to enhance the predictive treatment efficacy power of O BPIt will lead to incorporating the patient responsesuccess failurefrom past treatments to rank the efficacy of candidate treatments identified by O BP for HTN medication optimizationClinical resourcese gElectronic Health RecordEHRprovide incomplete patient informationinsufficient to build thememoryfunctionality using retrospective datasetsThereforein Aima clinical trialnwill be conducted to collect complete patient datasets that will be used to build thememoryfunctionaBP data from remotely monitored BPbEHR updates to the patientandapos s profile since last cycle data dumpandcpersonalized online surveys for patient reporting of medication adherencesideeffectsSEsand cardiovascular symptomsAimwill utilize the datasets to develop an efficacy proximity mapx month of candidate treatments to past treatments using IBL theoryThe proximity map will be based on scored past treatment sefficacydefined as a combination of success in BP loweringreducing of SEsand targeting minimum number of drugs with maximum BP lowering powerThe treatment recommendation alert will be sent to the physicianfor considerationThis study design allows immediate clinical adjudication of the IBLmemorythrough physicianandapos s choice to implement reject the treatment action alertmade availablex month for each patientMemoryenhanced O BP is expected to lead to a systolic BPSBPreductionmmHg by the end of the scheduledmonths treatment optimization periodSuccessful development and validation will position thememoryenhanced O BP for deployment at UCSF MC that serves approximatelypatients with an SBP andgtmmHgpoorly controlledand at high risk for primary or recurrent strokeheart failure or myocardial infarctionOur long term objective is to transform the reactive and punctuated nature of HTN medication treatment management into a proactive and ongoing component of patient care Our goal is to simplify the medication management of high blood pressureBPfor the treating physician while improving the quality of care of patients and reducing the associated costsWe will develop an artificial processing intelligence that will allow our current productoptima BPto enhance its efficacy evaluation of candidate medication treatments as a function of past treatments performanceoptima BP is a decision support clinical reasoning artificial intelligence designed to safely optimize medication treatment by a physicianindependent of in office visits in patients with poorly controlled BP