Electronic Health Reporter

Guest post by Edgar Wilson.

Sherlock Holmes famously captured the popular imagination with his uncanny ability to make wild, but accurate, leaps of logic to solve mysteries. By observing Dr. Watson’s suit jacket sleeve, upon their first encounter, he was able to deduce that Watson was in fact a surgeon, in the British Army, and had recently returned from Afghanistan, where he had sustained an injury.

When he slowed down to explain his reasoning, it was easy to follow; what made his deductions impressive was how quickly he would skip from observation to conclusion. I’m no Sherlock Holmes, but it seems to me that chatbots are poised to take over much of modern healthcare.

Following the Clues: The Chatbots Did It

The reasoning goes something like this: As we’ve adopted EHRs across a dominant share of the healthcare system, a parallel push for greater patient engagement and data transparency has led to widespread use of patient portals. So, EHRs lead to patient portals.

As more data is moved to portals through EHRs and digital documentation, there is increased patient interest in and demand for other digital and remote encounters and health resources. This, along with improving technology and competitive solutions, is helping increase adoption of telehealth. So, patient portals lead to increased telehealth adoption.

Finally, although part of the premise and value of telehealth is enabling face-to-face encounters between caregivers and patients without respect to geography, hospital waiting rooms, or other physical barriers, it changes certain expectations. Like all mobile and web-based services, telehealth feeds a consumer mindset that expects everything on-demand, all but instantaneously, and highly customized at that.

While portable patient records facilitated by EHRs and interoperability can help this, customization and on-demand healthcare doesn’t just put pressure on records and data. Patients want fast and personalized answers. As customer service centers, tech support, banks and virtually every other consumer-facing industry has learned, a lot of the on-demand load can be pushed onto increasingly sophisticated chatbots.

So, telehealth leads to growing expectations for on-demand clinical encounters and chat, which is provided by chatbots.

The Case for Chatbots

Retail has previewed much for healthcare: See how customer service upgrades have turned everyone into “The Most Important Person Here” wherever they go, in person or online. Consumers demand personalization, expedition, authenticity and they want it all exactly when and where they want it. And now, see how AI is not yet taking over the world, but is making FAQs and other routine customer service interactions painless for those answering, and interactive enough for those asking.

Retail is even making inroads to healthcare, as consumer-facing devices promise to measure and track all manner of health metrics. Statistics-loving sports fans witness the increasing digitization and quantification of athletes, games, injuries and training, and they want a similar level of insight and precision for their own care. Mobile technology is redefining and disrupting even the oldest and most stable of markets and industries, bit by literal bit.

So how long until the dry, repetitive questions doctors routinely must answer in check-ups and physicals are ethically and effectively offloaded onto chatbots programmed to triage and educate patients without wasting valuable human resources? How long until using telehealth to keep nonemergency patients out of the emergency room merges with using chat and AI — the basic recipe for chatbots — to keep healthy but curious or concerned patients from wasting time and money going through full encounters simply to get their general questions answered?

It doesn’t take a lot of sophistication to realize the benefits of AI at scale. Google has all but taken over the modern world by connecting searchers with answers to their questions; Wikipedia has all but bankrupted the encyclopedia industry with free, accessible, general knowledge. In a world where health literacy is so lacking in the majority of the population, some interactive resources could go a long way to chipping away at ER overuse and healthcare overconsumption, just by giving people an alternative to seeing the doctor.

Automation of Care, Automation of Crime

As quickly as potential benefits can scale, very real risks and both moral and financial hazards scale even quicker.

The growing popularity and implementation of chatbots has given hackers and cybercriminals a new way to scam, defraud, and generally abuse unwitting consumers. Sometimes that means hackers take over a company’s chat system with their own bot and solicit data. Sometimes fraudsters attract visitors with a spoof website, then use a bot to similarly extract volunteered data at scale from misled visitors. However it is done, it scales almost as well as a more conventional data breach, and can be harder to detect or track.

Healthcare is no stranger to the risks of new technology; rather, healthcare has become a favorite target of hackers and cybercriminals, seemingly incapable of keeping its newly digitized data secure. Throwing yet another pathway into the mix might seem destined to compound the problem, and further disintegrate patients’ trust and engagement.

As with the EHRs, so with the chatbots: it doesn’t have to be this way. Given the unsustainability of recurring hacks and data breaches, it seems clear that we can’t remain negligent in the face of new security and privacy risks brought on by new technology. The calculus (however flawed) that brought us into the era of EHRs determined that benefits warranted the change, even as the system struggled to prepare, implement, and learn to utilize digital records. A similar game of catch-up may well accompany chatbots in patient portals and telehealth engines: the potential cost-savings, in spite of the security risks to patients, may well justify offloading even a small share of caregiving to AI, rather than human providers.

Sherlock seldom left a case unsolved once he took it on. The case of the caregiver chatbots, it seems, must remain open for some unpredictable span of time before we know what role the tech will play in healthcare. For my part, I’m betting that role won’t be “none.”