(US) 1.800.388.4264 | (CANADA) 1.866.771.8231

Taking the guesswork out of excellence

In a recent article published in MIT Technology Review, it was revealed that a rather shockingly large number of AI-driven tools constructed to help clinical care teams combat COVID-19 were, to put it simply, failures. Of 232 predictive tools intended to either help diagnose patients or predict how sick they might become, none of them proved fit for clinical use. Another study referenced in the MIT article studied 415 tools, with none of them being deemed fit for use.

As the article states, “In the end, many hundreds of predictive tools were developed. None of them made a real difference, and some were potentially harmful.” These findings demonstrate an obvious failure to connect well-intended AI tools with real-life applications. Intelligent hammers were produced, only to be used on the thumbs of those who picked them up and put them into practice.

It is an unsettling revelation, albeit one that’s likely to put machine-learning developers one step closer to implementing tomorrow’s tools of efficiency on their next go-around. Until then, health systems would benefit from utilizing tools proven to help improve clinical outcomes and eliminate much of the guesswork that goes into programming appropriate inputs into any AI-driven tool.

As readmission rates and post-discharge needs rise across the country, NRC Health recognizes that patients, now more than ever before, need intelligent interventions to ensure healthier outcomes. Various AI-driven and predictive risk-assessment stratification models are often implemented in pockets, but honest assessments from experts throughout the industry should acknowledge that consistency is lacking and even a fully committed plan remains vulnerable to risk in relation to patients who fail to meet the agreed-upon criteria.

Patients who stay three days rather than four, who rely on urgent-care units instead of emergency departments, or who are otherwise healthy but incredibly forgetful might slip through the cracks when it comes to appropriate post-discharge care. They’re harder to identify through AI tools—but a standardized approach to following up with all patients can still give every one of them the opportunity to identify key risk factors that impact their health outcomes. Rather than rely on unconvincing models with inflexible or incorrect parameters, perhaps a better scenario is to keep it simple and take a holistic approach to care by letting the patients themselves—who, after all, know their bodies best—help you identify risk.

Leveraging technology, timing, and a clinically focused short question set that gets to the heart of highly correlated follow-up needs, NRC Health’s Care Transitions solution automatically makes post-discharge phone calls to 100% of patients to find the 17% (on the national average) who want or need to hear from caregivers. Eliminating assumptions and parameters, every patient is afforded the opportunity to reconnect with staff to address the risk concerns that leave people with worsening conditions, at risk of readmission, and/or incapable of full recovery.

NRC Health’s Care Transitions solution has been shown to positively impact all patient classes, proving useful outside of the theoretical realm and well into the practical. In a study of 18,788 heart-failure patients, patients who completed the discharge phone call (participants) experienced 29% better 30-day readmission rates than patients who did not accept the discharge call (non-particpants). Of 25,527 sepsis patients, participants experienced 19% better 30-day readmission rates than non-participants. Of 6,055 relatively lower-risk spinal procedure patients, participants experienced an incredible 63% improvement on 30-day readmission rates over non-participants. In other words, NRC Health’s Care Transitions solution made a real difference.

To date and at scale, there is nothing better at predicting the course a patient may take than that patient themselves. While AI-driven tools and technologies will undoubtedly have a place in healthcare, we must never forget the role of the individual human being and the power they possess to improve our understanding about how they should be treated.

In light of the fascinating article in MIT Technology Review and the realization of the ongoing disconnect between technology and the people it is intended to benefit, we strive to continue our pursuit of meaningful connections through Human Understanding.