Computers Will Not Replace Us

A 2017 study from the Massachusetts General Hospital and the Massachusetts Institute of Technology showed that an artificial intelligence system was equal to or better than radiologists at reading mammograms for high-risk cancer lesions needing surgery.(1) A year earlier, Google reported that computers are as successful as ophthalmologists in examining the retinas of patients with diabetes.(2)

Now the question arises, will computers replace physicians? The answer is “no”—because artificial intelligence (AI) is only as good as the humans programming it and the system in which it operates.

AI is already being used in the nonclinical aspects of healthcare. AI systems are able to analyze enormous amounts of data. Advances in computing power have enabled the creation and cost-effective analysis of large data sets of payer claims, EMR data, medical images, genetic data, laboratory data, prescription data, clinical emails, and patient demographic information.

COVID-19 has uncovered a health disparity between whites and non-whites that has been under the healthcare radar for decades. According to a 2017 report by the National Academy of Medicine on health care disparities, non-whites continue to experience worse outcomes for infant mortality, obesity, heart disease, cancer, stroke, HIV/AIDS, and overall mortality.(3)

Another major challenge is that many clinicians make their medical decisions that are not neatly documented as data points in an EMR. Clinical judgment at this time is not replaced by data and algorithms.

Some proponents claim that AI will make physicians as obsolete as the assembly line workers manufacturing automobiles, as bricks and mortar bookstores, and as writing checks and performing other financial transactions in a bank building. In 2016, Obermeyer and Emmanuel(4) predicted that machine learning will displace much of the work of radiologists and anatomical pathologists because AI will soon exceed human accuracy.

Artificial intelligence could be particularly beneficial in areas with limited access to healthcare.

So where will AI fill in the gaps in the current healthcare system? AI could be particularly beneficial in areas with limited access to healthcare.

Researchers at Stanford University believe their skin cancer algorithm could work on a smartphone, allowing people to screen themselves.(5)

And algorithms can handle a larger workload than humans.(6) For example, a dermatologist may look at only 12,000 lesions in his or her professional lifetime, while the Google skin cancer program has reviewed over 130,000 images.(7)

Also, computers don’t get tired. They can work 24/7; they can work much faster than their human counterparts; they don’t take sick days; they don’t burn out; they don’t require benefits, and they don’t argue with the office manager or the physicians.

Why the Computer Will Never Replace the Doctor

The healing relationship between a doctor and a patient has been based for 2500 years on collecting information regarding the patient’s history, including the environmental history, the social history, and the family history; conducting the physical examination; and, more recently, requesting laboratory tests or imaging studies to narrow the diagnosis and provide a hierarchy of diagnostic options or differential diagnosis. The doctor mentally ranks or prioritizes the options and only then initiates treatment for the patient. This method has worked for several millennia, but today that paradigm is no longer going to be what patients are demanding of the healthcare industry. One of the most recent advances to occur during the COVID-19 pandemic is the use of telemedicine to connect with patients. Now it is acceptable and even safe to provide care in certain instances without examining the patient.

An AI-powered diagnostic program can collate and summarize for the doctor tens of millions of similar cases, including the most recent publications in thousands of medical journals, and make it possible for the clinician to spot the zebras amongst the horses. The AI tool can make any doctor a super diagnostician without having to send the patient to advanced tertiary centers such as the Mayo Clinic or Cleveland Clinic to obtain a diagnosis, even when that patient has a rare disease or condition. AI will be an adjunct for physicians and will offer all available knowledge to recommend the best treatment option but still allow doctors using instinct, skills, and experience to override the AI evaluation.

No computer is going to be programmed to offer the human touch that only doctors can provide.

What’s the upside of AI? With AI crunching millions and billions of bytes of data, the doctor, nurses, and caregivers can focus on human tasks that no computer can accomplish, such as providing compassion, understanding, and empathy. No computer is going to be programmed to offer the human touch that only doctors can provide.

I am reminded of a true story. A patient in England was scheduled for elective surgery, which was postponed because of a shortage of operating room nurses. The family, which included an American doctor, was never informed about the cancellation. The surgeon walked past the family without offering an explanation or consoling the family. This English surgeon clearly lacked empathy and understanding about the needs of his patient and the patient’s family.

No computer will ever be able to replace this important human skill, that was unfortunately lacking in the anecdote above. Emotional and empathy training should be a part of every doctor’s medical school education as these are skills that, at the present time, cannot be accomplished by a computer.

AI is likely to surpass doctors in their ability to diagnose medical problems and recommend the best treatment for the patient. Although some thinkers predict that AI will eventually replace doctors, the evidence does not support this contention. Patients don’t want to be treated by a machine that delivers a hierarchy of diagnoses and treatment options. No patient wants to hear, “You have prostate cancer that has spread to your bones, and there is a 40% chance of you dying in five years.” Instead, patients want a compassionate and caring doctor to deliver that kind of news accompanied by a gentle pat on the back and reassure the patient that everything will be done to control the spread of the tumor and that the patient will remain comfortable and have minimal pain.

Our prediction is that this humanistic approach is going to give rise to a whole new kind of doctor. These new doctors will be a combination of nurse, medical assistant, social worker, and even psychiatrist. These compassionate caregivers will be trained to have the best communication skills, be readily available to console patients during times of trauma, and emotionally support patients throughout their treatment. The doctor could reassure a patient that the doctor had patients with a similar diagnosis and was able to arrive at a successful treatment and outcome.

These new compassionate doctors would not compete with computers in their ability to memorize minutiae such as the origin, insertion, and innervation of each muscle in the human body or how many ATPs are generated by the Krebs cycle. These new caregivers would be well trained, with skills requiring more emotional intelligence, not as a reservoir of countless facts and figures. These new caregivers would be an adjunct to the computer giving patients enhanced accuracy in the diagnosis and treatment of medical conditions. We see a future for medicine that would be kinder and gentler and a little more caring and loving.

The American Medical Association predicts a shortage of physicians in the near future, because the baby boomers are aging and retiring, with no increase in the number of medical students, compounded by many middle-aged and older doctors retiring earlier than in the past. With a relative scarcity of trained physicians, the laws of supply and demand will increase the cost of healthcare and perhaps drive down the quality of care delivered. Really, how can a doctor possibly manage a patient with multiple problems in the traditional 15-minute appointment? This scenario has the potential of rationing care and decreasing the amount of time that each doctor spends with a patient. (The exception is concierge medicine, where doctors limit their panel of patients and have the luxury of spending more time with each patient.)

The computer will not replace the doctor, but home health aides and personal care aides are going to be some of the fastest-growing professions in the United States. There is an anticipated growth of 1.2 million new jobs in the healthcare sector by 2026, according to the U.S. Bureau of Labor Statistics. At current figures, these caregivers will have an annual salary of $35,000, which means there will be more caregivers than physicians and at a reduced cost. Grand Aides (www.Grand-Aides.com), now operational in 61 programs in the United States including Accountable Care Organizations, commercial insurers, health plans, third-party administrators, employers, Medicare Advantage, Medicaid, and several Veterans Administration hospitals, has resulted in improvement in care and reducing costs of healthcare.(8) These caregivers are loving and compassionate and meet the needs of patients with chronic conditions who have limited activity and face difficulty going to the doctor’s office or urgent care centers. If history is to repeat itself, the industrial revolution and the information revolution of the 1990s resulted in an increase in jobs rather than displacing people from blue-collar jobs. In the long run, technology leads to more jobs and greater prosperity for all. AI is not going to be an exception, and healthcare providers have nothing to fear.

One of the most important requirements of the large number of patients who are being cared for in their homes, more than anything else, is human contact. This need cannot be met by a computer. Only the touch of another person can create and share compassion and empathy. For the foreseeable future, there is no possibility of computers that are able to feel emotion. It is in the unique ability of humans to provide this to each other that the computer will not replace us (physicians). In the current COVID-19 environment, where many Americans have been vaccinated, but social distancing is still recommended to control the spread of the coronavirus, we have seen how the absence of touch is creating a void in the human spirit and the socialization that is so necessary for happiness and good health.

Bottom Line: Physicians have nothing to fear about being replaced by AI. 19th-century weavers in England smashed the new industrial looms that they blamed for taking their jobs. They feared being replaced by technology, with the resultant loss of employment. This fear proved to be unfounded—industrialization proceeded, and the number of jobs increased, and so did the quality of life for the majority of the English population. We don’t foresee a day when doctors will start destroying computers fearing a loss of their jobs. We do see a time when computers will change how medicine is practiced and computers will enhance healthcare and improve the quality of care that we offer our patients.

References

1. Choy G, Khalilzadeh O, Michalski M, et al. Current applications and future impact of machine learning in radiology. Radiology. 2018;288:318-328.

2. Gulshan V, Peng L, Coram, M., et al. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA. 2016;316:2402-2410.

3. Dzau VJ, McClellan MB, McGinnis JM, et al. Vital directions for health and health care: priorities from a National Academy of Medicine initiative. JAMA. 2017;317:1461-1470.

4. Obermeyer A, Emanuel E. Predicting the future-big data, machine learning, and clinical medicine. N Engl J Med. 2016;375:1216-1219

5. Esteva A, Kuprel B Novoa RA, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542(7639):​115-118.

6. Brady AP. Error and discrepancy in radiology: inevitable or avoidable? Insights into imaging. 2017;8(1), 171-182.

7. Esteva A, Kuprel B, Novoa R, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542: 115–118.

8. Baum N. Grand-Aides: lowering healthcare costs and improving outcomes. J Med Pract Manage. 2021;36:186-188.