Hasta geri bildirimi epilepsi izleme sisteminin yanlış alarmlarını azaltıyor

Continuous EEG monitors collect far more information than patients usually receive, while their algorithms can mistake harmless brain activity for seizures. An NIH-funded team designed a conversational interface to create a two-way exchange between people and their monitoring devices. In a hospital study, 13 patients used a smartphone to ask about their physiological data and respond to real-time seizure alerts by marking them true, false or uncertain. The participants and system exchanged more than 1,300 messages, and over 92% of questions about brain activity received accurate, relevant answers. More importantly, responses to alerts allowed the detection model to fine-tune itself. Across the study, patient feedback reduced false alarms by about 80%; expert feedback produced a further small reduction. The result remains a proof of concept because testing involved few participants in a controlled hospital setting. Researchers therefore plan larger outpatient studies and are exploring a similar interface for deep-brain stimulation devices used in Parkinson’s disease.