LATEST

In a landmark 2016 study Johns Hopkins researchers estimated that more than 250,000 Americans die each year from treatment-related mistakes, making medical error the third-leading cause of death in the United States. As a former military flight surgeon trained in aviation accident investigations, I know well the hazards of misusing or mistrusting instruments.

In contrast to health care, aviation has been an early adopter of decision-support technology — the “fly by wire” flight-control computers that prevent unsafe operation of the aircraft and reduce maintenance costs and pilot workloads. Due to the progressive digitization of the cockpit and pilot decision support, flying by and trusting instruments is now essential for avoiding accidents. The U.S. Department of Defense’s new F-35 aircraft is so advanced that the pilot interacts continuously through a “heads-up” digital display projected on the helmet, providing total situational awareness. Pilots who aren’t adept at working with computer interfaces and don’t trust algorithms to help fly the aircraft will not just perform poorly, they’ll crash on takeoff.

Though health care is still in the early stages of adopting AI and digital technologies, it is already making great strides. In radiology such technology will augment human competence in image recognition. Digital technology now assists with detection of diabetic retinopathy, heart arrhythmias, and dermatologic diseases. Diabetes care will be enhanced through wearable sensors and AI to detect or prevent harmful hypoglycemic episodes. Remote monitoring will prevent or detect deterioration in chronic conditions such as asthma, chronic obstructive pulmonary disease, and congestive heart failure, reducing costly hospital re-admissions. And technology has the potential to connect medical professionals across the continuum of care, reducing the potential for patients to fall through the cracks.

But to realize the full potential of AI and other digital technologies we will need to overhaul medical education for future physicians and nurses and rethink professional development for current caregivers.

In medical education today, students acquire medical knowledge via memorization and gain experience and good judgment through lengthy apprenticeships. Their degree of competence as physicians will depend on what information they retain and what clinical experiences they are exposed to in the course of their careers. Professional development often consists of training provided by medical device manufacturers and pharma reps, which leads to highly variable levels of proficiency among caregivers. And gaining proficiency often means practicing on patients.

Professional development, like education, needs to incorporate more digital technologies and simulation platforms to reduce variability in competence and experiences. And caregivers must learn to trust AI and digital technologies to augment their clinical judgment. Every year there appear more predictive models capable of foreseeing outcomes and complications of treatments, practices, medications, and procedures. Yet, as anyone who routinely attends mortality and morbidity conferences in hospitals knows, there is the recurring problem of individual clinicians overriding available specific patient data or altering proper care as a result of confirmation bias, leading to medical errors.

By better teaching students and practitioners how to fly by instruments and to trust those instruments, we could both improve training and solve many of the problems that lead to those medical errors:

• In medical education and professional development, cognitive computing and robust simulation platforms can shorten training times and provide safer environments for learners to make mistakes. Simulation platforms allow better training to standards and can be adjusted to the learner’s pace of learning. They also allow in-depth training in tasks, demonstration of cognitive knowledge, exercise of judgment, and improvement in team dynamics.
• Digital health assistants (also known as digital coaches and virtual assistants) can help doctors take a more complete patient history and augment patient education.
• AI-enabled technologies can help provide more accurate diagnoses earlier (studies have shown 20% to 30% of initial diagnoses are wrong) and provide more accurate treatment plans.

AI technologies and digitization will also be critical for helping create the “learning health system” envisioned in the federal “Precision Medicine Initiative” inaugurated in 2015 (and now known as “All of Us”). Prompted by that initiative, the leaders of the federal health agencies (of which I was one as assistant secretary of defense for health affairs) embarked upon an unprecedented collaboration in bio-medical research to transform scientific evidence to support health care decisions and public health. The initiative includes assessing the landscape of data available to clinicians and patients for use in clinical decision making, identifying opportunities for enhancing the available body of clinical evidence, and initiating demonstration projects that use digital data from government sources and payers, providers, and patients. The work that the federal leaders in health care have undertaken will help produce the instruments that providers and health systems can use to improve effectiveness and efficiency of care:

• Neural networks will be able to mine medical records, design personalized treatment plans, accelerate new drug discovery, and help mitigate social determinants — like housing, education, availability of nutritious food, neighborhood violence, and more — of adverse health outcomes.
• The virtual-reality and gaming technology will provide better and more efficient skills development and will also improve patient engagement in their own well-being.
• Wearable sensors will provide the platform for connected health systems and proactive management of chronic disease to prevent deterioration.
• The “internet of medical things,” connecting health care providers with medical monitoring devices, could make the home the front line in health care, where chemotherapy, dialysis, antibiotics, and other treatments could safely be delivered, using AI and digital decision-support technologies.
• Enhanced data analytics and modeling will open opportunities for personalized health-management strategies and better population-health management.

As the authors of the Johns Hopkins study pointed out, medical errors aren’t due to inherently bad doctors but often due to systemic problems, including poorly coordinated care, fragmented insurance networks, the absence or underuse of safety nets, and other protocols, in addition to unwarranted variation in physician practice patterns that lack accountability. AI and digitization hold out the promise of addressing those problems — pulling together fragmented networks, coordinating care, standardizing protocols and practices, and providing the ultimate safety net in “fly by instruments” decision support.

from HBR.org http://ift.tt/2FCBXET