Empathy, data science and better healthcare.
Springtime in Norfolk, Virgina is grey. The city is surrounded by water on 3 sides, being at the confluence where the James River meets the Chesapeake Bay. Grey fog permeates everything. The average spring evening temperature is around 46 F. That evening we were badly in need of some warmth. I had come to Norfolk for a company site visit with Horst. As we stepped into the dining room, there was a fire crackling. That orange glow, the crackling hisses and pops, and the warmth. It was just what we needed.
“So, David, let me tell you about a computer program I created,” Horst said, as we settled into the leather chairs.
At the age of 14, he had left home to go to a hotel training school, which in those days meant that you worked long hours in a hotel with little or no pay. Somehow, he worked his way up, and years later, Horst Schulze founded the modern iteration of the Ritz-Carlton. Today, he is widely regarded as the father of the modern hospitality industry. If someone knows how to deliver an exceptional consumer experience, it’s Horst.
But computer programs? I was intrigued.
“So, what I did was create a database and had the housekeeping staff log each time they replace the batteries in the TV remotes and change the light bulbs in the rooms.”
“Then I extrapolated when the batteries and light bulbs were likely to fail based on their rated capacity and life.”
“And just before each of them would have failed, the program would send out alerts to have the housekeeping staff go and change them,” Horst explained to me.
“So this way, no one ever will experience a burnt out light bulb or a dead remote control,” he concluded. “Because I know how that annoys everyone.”
That’s 27,650 rooms, 55,300 remotes and around half a million light bulbs. Roughly equivalent to the entire town of Medford, Massachusetts.
I was blown away by why he wanted to solve this and how he did it. After some reflection, I realized that it came down to this: It was driven by empathy, codified by data science and delivered as just-in-time insightful help (actually, right before it’s needed), a la Ritz.
Primatologist Frans de Waal, in his book “The Age of Empathy”, provides a useful framework to understand how empathy can lead to insightful help. De Waal describes 3 necessary components: visual contact, group identification, and non-competition. Let’s consider them in the context of Healthcare and AI.
- Visual contact:
You can’t empathize with and help those whom you can’t see. De Waal refers to this as establishing visual contact between the ‘empathizer’ and the ‘empathizee’. In Healthcare, this would be between the provider and the patient respectively. I can’t help but wonder if there is a deeper lesson here for Healthcare. As discussed in my previous article, there is an increasing interest (and capability) in the use of AI to derive contextual meaning through the vectorization of contextual features. The average medical record is around 200 pages. A physician typically has around 15 minutes to review. As you can imagine, AI can become quite useful to clue you in to what is pertinent and important, so that you can get the complete picture and “see” the patient. Would this lead to better care? Absolutely.
2. Group identification:
This refers to us being more likely to empathize with those whom we feel are similar to us in some way. They are within our group, somehow. This presents an interesting opportunity. Providers are humans with our own biases. I wonder would having an AI match and highlight characteristics about the patient that are important to the provider elicit greater empathy? For example, would knowing that an elderly patient was a Korean War veteran (like my grandfather) make me more empathetic towards that patient? Would that patient receive better care as a result? I would like to think of myself as being fair and impartial, but to not at least consider this, would not be fair and impartial.
From the patient’s perspective, the potentials of using group identification through the use of AI to improve treatments are tremendous. For example, we know that one of the strongest motivators towards healthy behaviors is having a loved one counsel you towards it. To that end, there are programs and apps that allow grandchildren and other loved ones to record motivational video messages to help seniors quit smoking. Photo-realistic virtual video can now easily be generated through the use of Generative Adversarial Networks (GANS) using relatively modest hardware. It is not hard to imagine apps generating dynamic videos in the likeness of loved ones (with the proper permissions of course!) to motivate patients towards healthy choices in the near future.
3. Non-competition:
This last one is perhaps the most interesting of all. In some sense, it’s obvious. It would be difficult for us to empathize and help if we perceive someone as a competitor of ours. Within Healthcare, competition often manifests itself as the power struggle over knowledge between the provider and patient. Because of years of study, training and experience, a provider usually maintains an asymmetric power of knowledge over patients. Many providers use this advantage as a way of feeling safe about their job security. After all, if the patients know what I know, wouldn’t I be out of a job?
I would contend that this is in fact an erroneous view. Just because I understand theoretically how oil changes are done, and can decipher what the service manager is trying to tell me, doesn’t mean that I would want to do them myself (or that I am qualified to do so).
Having providers and patients speak the same language and having some shared knowledge base can build trust and rapport. If the barrier to entry is that medical language is obtuse then we already know how to solve that with AI. The reason that modern language translators work so well, is that they use Long Short-Term Memory models (LSTMs) not to translate word for word, but to associate a phrase in one language with another phrase in another language. Isn’t that exactly what’s needed between “Medicalese” and English?
Like that crackling fire in grey Norfolk, delivering exactly what patients need with empathy is what all providers should aspire to do. Doing so will result in better care. AI can help.
“Solving problems in Healthcare with AI”:
1. “3 Problems”: https://lnkd.in/eHBZdJQ
2. “The Inheritance”: https://lnkd.in/eB6qQPM
3. “Empathy”(this article): https://lnkd.in/ek7ynM3
4. “One Trillion Dollars”: https://lnkd.in/eGX3Cgb
