Telemedicine, by using advanced sensors and transmission technologies, helps health care personnel to diagnose and treat patients remotely. National Taiwan University Hospital established its telehealth care center in 2009. Over 4000 patients have attended this program. Most of these patients have cardiovascular disease. The case managers of the telehealth care center work around the clock to contact and monitor patients. These telephone contacts had been stored in the center. Over 2000 hours or 100GB voice records can be generated in one year. The case managers play important role in the care program. If there is an algorithm to help case managers catching the main problem of a patient on the telephone, triaging the patient and showing relevant lab data on the screen, the loading of case managers can be relieved greatly.
The goal of this project is to use our long-term telephone records to convert manually speech into text, highlight the keywords, retrieve the speech intention, classify the clinical problems, and facilitate the daily service. An explainable AI will be emphasized in this project, to ensure human adoption of AI-based clinical recommendation.
The potential applications of this project include AI-facilitated problem classification, AI-based prognostication, and identification of clinical conditions. These techniques can be integrated to a single platform to establish an AI-based telehealth case management system. By the help of this AI-based system, the case manager can deal with much more patients at the same time. The individual technique of this project can also be incorporated into the conventional outpatient or inpatient health care scenario to help the nursing staff to manage the demand of patients.