Cognitive Technologies Will Reduce Hospital Readmissions

Published on: 

Hospitals are facing hundreds of millions in penalties for readmissions, and they are trying coding gimmicks to avoid these. Instead, it's time send a nurse home with every discharge.

An article by Lucette Lagnado in Tuesday’s Wall Street Journal discussed the poor recall that patients have after discharge from the hospital and cited data that patients forget 40% to 80% of the information their doctors tell them immediately after the discussion. Lagnado went on to discuss a few innovative hospitals and their approach to addressing this issue through technology leading to decreased readmissions.

The next day in the Journal, the first page contained an article by Christopher Weaver, et al, that discussed how hospitals decrease re-admissions basically due to a coding changes that move inpatient admissions to "observational" status. The coding changes reduced the previously reported reduction in readmissions by 3/4.

About 49,000,000 people are currently receiving Medicare benefits. There are about 290 discharges each year for every 1000 people of Medicare age. This accounts for about 14 million hospitalizations each year. Medicare is assessing penalties (amounting to an estimated $428 million dollars in 2015) on congestive heart failure, heart attacks, pneumonia, chronic obstructive pulmonary disease, and elective hip/knee surgery readmissions. And those penalties are sure to rise.


Hospitals are desperate to reduce readmissions and both articles in the Journal contained ideas on how to reduce readmissions... but one resulted in improved care, the other only clouded the issue.

The use of video and audio recordings are useful tools, as the first article discussed, and may actually make a small difference, but there is a need for a more comprehensive approach—a cognitive approach—that not only assists with helping people understand the discharge instructions, but also teaches them about their condition, monitors adherence to the prescribed therapy, and produces a better outcome. This approach produces a sustainable reduction and improved overall care.

I propose that the next generation of solutions will incorporate cognitive solutions in the form of specialized virtual health assistants (VHA) residing on a smart device. A VHA, also termed an avatar, could not only trigger the video and audio recordings discussed above, on demand, but could also monitor a number of physiologic parameters such as salt intake, temperature, respiratory rate, ambulation, heart rate, glucometer results, weight, sleep quality, and medication adherence, to name a few.

By allowing the artificial intelligence/natural language powered VHA to access data from "wearables" and connected devices, as well as the hospital discharge instructions, prescription instructions, and prescribed lifestyle changes, the agent could, using motivational interviewing, encourage people with chronic illness recently discharged from the hospital to adhere with the numerous instructions and track, in real time, response to therapy as well as adherence to the treatment plan—basically creating a virtual nurse that accompanies each patient home.

Of course, the VHA could detect variance from prescribed therapy and escalate the issues to the hospital and physicians involved as needed and provide a video bridge to the caregivers as needed with a so-called virtual visit.

It seems counter intuitive that hospitals would walk away from what trends to become a half a billion dollars when an VHA could be built for the cost of just the salary of a handful of nurses. Once built, these agents are infinitely scalable and would continue to grow in knowledge, scope, breadth and utility.

The technology is ready and hospitals have a compelling need and have tried accounting gimmicks. Who will be the visionary to try a real solution?