Human Technology Frontiers

Learning systems for health care workers to perfect safety-critical work

Health care already makes up at 17.8% of GDP and is projected to become both the largest and fastest growing job sector in the next decade. However, the ratio of potential caregivers to seniors is expected to shrink from 7:1 in 2010 to 4:1 by 2030, significantly straining the ability to deliver quality health care. To address that issue, our team of CERC researchers aims to explore the implications of boosting worker productivity and satisfaction via intelligent automation in the physical activities of health care delivery in hospitals, where 6% of gross domestic product is consumed and patients are uniquely vulnerable to defects in clinician bedside activity.

Virtual assistants, or “ambient intelligent monitors” (AIMs), already use computer sensing algorithms in some work environments to continually discern physical activity, reduce worker cognitive and physical burden, and improve worker performance via behavioral prompts. Despite great promise, little is known about how computer assistants such as AIMs will change bedside care in hospitals, what forms of human technology interface clinical workers prefer, or how work environments can be redesigned to maximize their value to patients, families, and clinicians. 

Building on an ongoing pilot initiative that deployed computer-vision AIMs in hospitals, the Human Technology Frontiers project will address these critical issues. The project involves a tight collaboration among computer scientists, electrical engineers, clinicians, behavioral scientists, and health care economists across Stanfords schools of Engineering, Business and Medicine, as well as clinicians at Stanford and Intermountain hospitals. 

Real World Intelligent Automation Applications

Problem statement: failure to adhere to hand hygiene protocols contributes to the 1.7 million health care-associated infections (HCAIs) and to the 98,000 HCAI-related deaths annually

Application goals:

• Develop algorithms to recognize hand hygiene events

• Identify opportunities hand hygiene improvement

Projects and Outcomes

Human Technology Frontiers will establish foundational interdisciplinary knowledge about interdependencies among future technologies, future workers, and future work and for the real-world deployment of AIMs in diverse hospitals.

The research develops innovations in hospital-embedded hardware, software algorithms, and integration of AIM technology with workers and workflows. Assessment of clinical, behavioral, operational, and economic impacts will guide the research toward its high-level goals. Research activities will be integrated across disciplines to enable synchronous development of, and learning from, hospital applications of AIMs.

The research will proceed in cycles via multiple, parallel streams, with the initial focus on clinician hand hygiene, health-critical ICU care bundle actions, and laparoscopic surgeon skill. Each research cycle will include analysis, design, development, implementation, and evaluation for a proposed use of AIMs. One stream will develop hardware for capturing privacy-protected data on edge devices including RGB cameras, depth sensors, and microphones. A second stream will develop algorithms that discern success and failure in enacting intended clinician bedside activity. A third stream will develop ways for health care workers to interact with AIMs for both input and output, in order to perfect intended action. A fourth stream will measure the clinical, behavioral, operational, and economic impacts of the AIMs.

Project work streams and research methods

Applying computer vision to health care

Work Streams