Before a clinician can decide what should happen, the clinician must understand what is happening.
For a person with several chronic conditions, that understanding may require bringing together diagnoses, medications, symptoms, laboratory results, physiological measurements, functional changes, mental health, adherence, recent hospitalizations, specialist recommendations, social circumstances, and the person's own experience. Just as importantly, the clinician needs to understand how those things have changed over time and how they relate to one another.
Much of this information may exist somewhere in the healthcare system. The difficult part is assembling it into a coherent understanding of the person at the moment a decision needs to be made.
Human cognition is powerful—and selective
Clinical expertise is an extraordinary cognitive achievement. Experienced clinicians do not reason by consciously considering every available fact. They recognize patterns, organize information into meaningful structures, and draw on years of accumulated knowledge to focus attention on what is likely to matter.
That ability is essential because human working memory is limited. In his influential 1956 paper, The Magical Number Seven, Plus or Minus Two, psychologist George Miller explored limits on human information processing. Later research refined that understanding. In a widely cited 2001 review, cognitive scientist Nelson Cowan found evidence that the capacity of attention is closer to three to five independent chunks of information, averaging approximately four under conditions that limit rehearsal and grouping. Cowan, Behavioral and Brain Sciences, 2001.
Expertise helps enormously by allowing many observations to be organized into meaningful patterns, but it does not eliminate the underlying constraint. As the number of conditions, medications, observations, clinicians, and care settings increases, information must be selected, compressed, or left outside the immediate field of attention. Usually that works remarkably well. Sometimes, however, the thing left outside the field is the thing that matters.
The burden is reconstruction
Healthcare's information systems have become very good at preserving individual pieces of information. The medication is in the record. The hospitalization is in the record. The laboratory result is in the record. The specialist's note is in the record. The patient's report of worsening fatigue may be somewhere else in the record.
But storing information is not the same as maintaining an understanding of the person. When those pieces remain fragmented across encounters, notes, systems, and time, the clinician must reconstruct the context in which they become meaningful. What changed? When did it change? What else was happening? What was tried? What happened afterward? Is today's observation unusual for this person, or part of a pattern we have seen before?
We have designed healthcare information systems that require extraordinary human cognition merely to reconstruct enough context to exercise clinical judgment.
The more complex the patient, the greater that burden becomes.
Remembering and relating
There are two parts to the burden of reconstructing a patient. The first is remembering: preserving what has happened across encounters, conditions, medications, measurements, symptoms, interventions, and time. The second is relating: understanding how those things connect to one another.
Healthcare systems are reasonably good at storing information. But storage is not memory in the sense that matters here. Useful memory preserves not only an observation, but its place in the person's history: what was happening around it, what preceded it, what followed it, and how the person responded.
Relating is harder still because the significance of an observation often lies in its relationship to other observations. A three-pound weight gain may mean little by itself. Its significance may depend on whether the person has heart failure, whether kidney function has changed, whether a diuretic was recently adjusted, whether medication adherence has slipped, whether shortness of breath has increased, and whether a similar pattern preceded an earlier hospitalization.
Each observation may be easy to understand individually. The difficult task is remembering and maintaining the relationships among them across time. Simply giving clinicians more data cannot solve that problem; more information can increase the burden of reconstruction rather than reduce it.
This is where the capabilities of machines and humans can become complementary. A machine can remember far more of a person's evolving history than a clinician can reasonably keep in mind, and it can continuously maintain relationships among those observations. The purpose is not to present all of that information to the clinician. It is to use remembering and relating to make what matters easier to notice.
At Senscio, this is the purpose of HealthGraph™ and the Digital Twin for Health™. They are designed to preserve accumulated knowledge about the person and the relationships within it, so that today's observations can be understood in the context of what came before.
The clinician should not have to reconstruct the patient before treating the patient.
The machine can help remember and relate. The clinician can bring judgment to what emerges. That takes us directly back to the human-machine synergy of the previous essay: the machine changes what the clinician notices; the clinician decides its significance.