Patient using a virtual agent

Optimizing Hospital Workflow and Quality through Patient Engagement[span]Using wearable accelerometers to detect sleep states for a personal health informatics interface[/span]

NEU mHealth Group Investigators[span]Aida Ahyaei, Stephen Intille[/span]

Collaborators[span]Timothy Bickmore (PI), Harriet Fell (Northeastern)[/span]

Sponsors[span]This work is sponsored by CIMIT[/span]

Can sleep states be detected well-enough from wearable accelerometers to provide effective input to a virtual character designed to be a personal health guide for a hospital patient and improve hospital workflow as a result?

We propose to improve inpatient care workflow through the development of a revolutionary patient-facing technology platform, the “Hospital Buddy”. The Hospital Buddy uses an intuitive, conversational computer interface that has been successfully evaluated in 10 clinical trials involving over a thousand patients and consumers of all levels of health and computer literacy. The Hospital Buddy is an adaptation of this technology that provides continual health counseling during a patient’s hospital stay via a bedside conversational agent equipped with sensors (RFID tags on staff, accelerometer on patient, microphone next to bed) that can detect certain events occurring in the hospital room. The Hospital Buddy talks using synthetic speech and animation and the patients respond using a touch screen attached to an articulated arm that can be positioned in front of them while they are in their hospital beds (this bedside platform has been used in two clinical trials at Boston Medical Center/BMC). In the proposed effort we will extend our existing platform with a suite of sensors as well as dialogue content and logic to empower patients, improve patient safety, and improve hospital staff efficiency. Specifically, the system will: (a) track and promote sleep, establish and track patient’s questions and issues, and monitor patient’s symptoms (e.g., pain) to ensure that patients are as alert, engaged and informed as possible about their care, leading to more efficient, productive, and error-free patient-provider interactions; (b) reduce time to respond to alarms (i.e., nurse call button, infusion pump, pulse oximeter, bed alarm, etc.,) and reduce false alarm rates by providing data on response times to the patient and providers for discussion, and counseling to the patient on how to reduce false alarms (e.g., due to motion); and (c) automatically track all patient-provider interactions (via sensor logs) and elicit patient feedback on these interactions, providing data that can lead to systemic changes to improve hospital workflow. In addition to these primary workflow-related outcomes, the Hospital Buddy will significantly increase inpatient satisfaction and provide a research testbed that can be used to explore a wide range of patient-facing interventions.

Representative publications:

None yet. Project underway Fall 2012.

Comments are closed