Making Medicine Human Again
The launch of Chat GPT in November 2022 brought artificial intelligence into the public conversation, provoking speculation about everything from revolutionary assistive technologies to the wholesale replacement of human labor. The truth, as always, lies somewhere in the middle. When it comes to the medical field, many, like Dr. Eric J. Topol and endocrinologist Aaron B. Neinstein, see AI as a powerful tool that can help to make medicine even more human, empower patients to take charge of their own care, and free practitioners to be healers again.
How AI Can Augment Human Medical Practice
There are a number of ways that artificial intelligence – Machine Learning (ML), Deep Learning (DL), and Large Language Model (LLM) technology – can make the practice of medicine easier, faster, and more accurate.
Automation of Time-Consuming Tasks
A study published in the Annals of Internal Medicine found that physicians spend an average of two hours doing paperwork for every four hours spent with patients. Doctors in another study by Dr. Christine Sinsky of the American Medical Association spent 27 percent of their time with patients and just over 49 percent on paperwork. And even in the examination room, nearly 53 percent of physicians’ time was spent filling out forms.
In addition to forms, administrative work can include catching up with patients’ medical histories, ordering tests and screenings, analyzing test results, reviewing medications and treatments, seeking insurance coverage for individual patients’ care needs, correspondence, and much more.
Large language models (LLMs) can lift a lot of this burden from practitioners’ shoulders. Set up correctly, AI can use the same information to populate forms for medical, insurance, and legal purposes. A practitioner can use an LLM like Chat GPT to generate correspondence and compile reports.
And, as with image analysis, a properly trained AI system can spot potential issues, such as medication interactions or insurance conflicts, which a human, overwhelmed with administrative work for multiple patients, may miss. It also has the potential to spot or even predict medical errors before they happen, which could save as many as 200,000 lives and $1.9 billion dollars per year.
Increased Accuracy and Efficiency
AI- based systems show incredible potential for not only analyzing large amounts of data quickly, but also making decisions based on these analyses. This ability is already being harnessed in image analysis, particularly in radiology. ML algorithms can analyze vast numbers of images far more quickly than a human practitioner can. In addition, AI can pick up on features and patterns that humans might miss. As a result, AI image analysis can help practitioners to make faster, more accurate diagnoses.
AI’s unparalleled prowess in pattern matching furthermore has the potential to perform more complex tasks with greater accuracy, improve clinical workflow, and improve treatment options. Although bias in training data has been shown to produce bias in AI output, carefully trained and monitored AI systems can reduce the danger of bias from and between human observers.
Empowering the Patient
Dr. Topol sees huge potential for AI systems in democratizing healthcare and empowering the patient.
Topol predicts that in the future, many diagnoses will be made as a result of patients capturing their own data through wearables, sensors, and home monitoring technology. Diagnoses of conditions that used to require a clinician, such as urinary tract infections, skin lesions and cancers, heart rhythm, and so forth, can originate from a patient’s own data.
Interoperable algorithm-driven devices are already helping patients to manage conditions like diabetes, as Dr. Neinstein has observed in his own practice. At one time, type 1 diabetic patients were issued a blood glucose monitor and an insulin pump and had to make their best guess how much insulin to instruct the pump to deliver at any given time, factoring in not only food, but sleep, activity, stress, and other things. Today, automated insulin delivery systems use continuous data from the glucose monitor to automatically adjust dosages, freeing the patient to go on about their lives, and freeing doctors to spend less time discussing insulin dosage calculations with their patients.
As Neinstein told Medscape, “My practice of helping people with type 1 diabetes didn’t disappear. It’s been transformed for the better. We’ve realized that I was a poor substitute for a computer.”
And, as well as an AI system can crunch data, it’s a poor substitute for human judgment, compassion, and a good bedside manner.
Concerns
Any time there are changes as potentially sweeping as those that future AI developments will bring, there will be concerns. Data security will be primary among these. Greater interoperability means sharing data across devices, and when data sharing between AI systems is unsupervised and instantaneous, care must be taken to safeguard patient privacy.
And while AI systems can be a revolutionary tool, it will be vital that human practitioners have the last word.
AI Can Free Practitioners to be Healers Again
Artificial intelligence systems are already revolutionizing medicine in numerous ways. The potential is there to increase not only efficiency, but also accuracy of diagnosis, as well as to individualize treatment. What’s more, freed from time-consuming data analysis and crushing administrative burdens, physicians, with the assistance of AI systems and devices, have the potential to practice the art of medicine again.