Medicine today is single-encounter oriented with documentation largely tailored for legal purposes rather than medical care. When an encounter ends, the physician must sign off the documentation and no further changes can be made; any corrections require another document that must also be signed off, but this is seldom done.

A further problem with current medicine is that many diagnoses are differential diagnoses instead of confirmed ones, which may be OK because many diagnoses require the same treatment. These incorrect diagnoses, together with the lack of corrections, may downgrade later medical care and also obscure later research using a patient’s medical information.

An additional major problem is the lack of a complete patient medical record, and perhaps many different ones in different medical organizations. A particular problem is small medical organizations which do not have robust automated medical record systems.

Whether or not there is a complete medical record, the one-encounter-at-a-time model fails to identify continuing care for the patient. This may not be a problem if a patient is seen within a relatively short period of time for an ongoing medical condition or if a physician is truly the patient’s primary physician. Otherwise, this one-encounter-at-a-time approach is a problem. There may be a lack of coordination, consistency, or continuity of care for a medical problem.

Disease histories today are embedded within encounter documents, often with the disease history tailored for the purpose of the specific medical complaint. (Instead, there could be separate longitudinal disease histories for each of a patient’s medical problems kept up to date by physicians and audited by physicians and patients for correctness; in this way, such disease histories could improve medical care and make later medical research easier.)

Today, physician proficiency in performance of procedures is not often tracked, and sometimes physicians with good bedside manner are esteemed by patients although they perform poorly on procedures.

Especially for inpatients, multiple physicians may be working together in the care of a patient (e.g. attending and specialty physicians). There should be better support for the communication and coordination between them.

Today, patients almost always come in for care when they have a medical problem or to be tested if they have a current medical problem. In the future there is likely to be more treatment of diseases before they occur; for example, today, potential future colon cancer can be treated by removing polyps found during a colonoscopy. In the ideal world, dementias would be treated and stopped before their devastating effects on patients.

Today, when a physician cares for a patient, he or she is concerned with future mortality and morbidity. A physician should also consider the patient’s future quality of life. Currently, medicine in the U.S. seldom measures a patient’s quality of life.

Outcomes of treatment are insufficiently recorded, so it may be hard to evaluate treatments. Biomarkers that could predict or evaluate outcomes are often not recorded.

Management of a patient’s medications is often inadequate, especially when there is no guaranteed complete patient medical record that shows a complete current list of a patient’s medications. For example, because of the lack of a complete list of a patient’s medications together with orders, drug interactions, drug allergies, or duplicate orders may be undetectable.

AI is currently insufficiently incorporated into medical care. And there are at least three concerns about AI:

• AI may determine diagnoses based upon physician expert opinions rather than actual outcomes.
• AI taking over diagnosis may cause physicians to lose their expertise in diagnosis.
• AI must be able to identify inconclusive results as well as negative and positive results.

Michael R. McGuire is the author of A Blueprint for Medicine.