Ogre in Disguise or Angel We Need in Healthcare?

Advances in AI and the platforms that support it have been speeding ahead at incredible increments, as we’ve witnessed by DeepSeek and various iterations of Anthropic AI. We must consider the potential benefits and unforeseen drawbacks of these advancements in healthcare.

To date, what we have been seeing in AI in healthcare is directed to managing data, data systems, and monetizing areas where potential is seen. This leaves us with one area that is of vital interest to all of us: we, the patients, need to understand how we are being considered. In other words, is healthcare and medicine, more directed toward management and giving less consideration to patients and their care in diagnosis and treatment?

As one physician recently told me, he is leaving medicine and doesn’t want anything to do with it in the future. In his words, he felt that private equity and AI were taking over healthcare, but not in a good way. “When I entered medicine,” he said, “we had all the patients’ information on one index card.”

At that point, he had a somewhat defeated look on his face, and he didn’t say anything more. Several months ago, another physician told me she felt what was being practiced now was “factory medicine.” What did she mean by that? She gave me several indications of how there was less care in selecting physicians for particular slots in practices, and she felt that too many were not adequately experienced or trained for what they were being asked to perform. All that was needed was an MD or DO, and they were hired for available slots.

These were only two examples, but I have others, too. Discouraged, another physician will leave the profession in two to three years. In fact, he could work at least two to three more decades in his specialty. However, a large hospital is pressuring him to sell or lose referrals. He is not selling and would rather close his practice. The loss here is obviously to the current patient base, who will have to seek care elsewhere.

The relationship and the experience that has been gained on both sides will also have been lost. I have also had to come to terms with how medicine is currently rotating physicians around to various offices, much to some of their dismay. As a result, I now have three new physicians with whom I have no prior contact. Do I need three? I doubt it.

What Does the Future Hold?

The further improvement of remote patient monitoring (RPM) will be one story of 2025. The pandemic hastened the adoption of RPM technologies, which are now fundamental to the treatment of chronic diseases, the recovery from surgery, and promoting healthy lifestyles.

There will be far-reaching effects on healthcare delivery as these technologies advance. This new technology allows physicians to track their patients’ vitals in real time, gather helpful information about their health, and act faster in the event of an emergency. Not only may this lead to better health outcomes for patients, but it could also help ease strain on already-strapped healthcare systems by preventing unnecessary visits to hospitals and emergency rooms.

But what of the lower-income patients who use hospital emergency rooms as their primary care site? If they can’t use RPM, what do they do? We know that the emergency room is a drain on finances, and charity care is even more of a drain. Hospitals are trying to decrease the patient load in those areas.

Does RPM truly represent an advance in medicine? Read carefully, and you will notice that physicians are becoming monitors rather than face-to-face providers. We are removing the human aspect and replacing it with technology.

Anyone who has ever studied psychology knows that human interaction plays a significant role in health and health maintenance. A digital device, whether implanted or worn on your wrist or hand, cannot replace this interaction, and we would have to question how sterile these environments will become. In fact, will they put physicians out of work?

Also, we need to question the infallibility of AI in terms of diagnostic competence and reasoning, as well as program bias. Because it’s computer-driven, it does not mean it is without fault, and therein lies a serious issue. No.

We’ve seen enough professional articles to know that not all technology is what it is advertised to be, such as robotic surgery or other forms of intervention. But patients can be blown away when they hear that instead of a real surgeon, they will have an automated piece of technology. Does that mean it’s better? No, unfortunately, it does not, but it may mean it’s more expensive. Once you buy equipment, you have to use it.

Even the Cleveland Clinic has indicated that proper standards and oversight are needed for all AI or it is inevitable that harm may result from it. But the advances, especially in specific radiology and breast cancer specialties, are impressive, and this hospital recognizes how it can contribute to healthcare.

One area where we may show some improvement is in note-taking. Some hospitals have a note-taker accompanying a physician on a patient’s visit. But there’s a new fillip in AI that addresses this.

Ambient listening, which uses machine learning to enhance audio, is a voice-recognition system that listens to and analyzes discussions between patients and providers in real-time, extracting pertinent information to be used in clinical notes and to meet billing and coding standards.

Organizations have examined and discovered obvious return on investment (ROI) around ambient listening solutions in terms of clinical efficiency and reducing burnout, which is why they are picking this as their first AI step. Does it address personnel burnout? This is a question that is still being posed in research.

Massive databases, including pictures, text, and other data required to train models, are the backbone of many AI systems. Despite the careful curation, some datasets may occasionally include illegal or unethical material.

In order to make sure that AI systems do not respond to users with damaging information, researchers implemented a mechanism called reinforcement learning from human feedback. To make AI systems more helpful and trustworthy, researchers use carefully selected databases of human preferences.

But if a researcher is unaware or insensitive to bias, can they truly curate AI? It has long been a concern that once a program is written and it absorbs scripts from previous programs, there is no way of knowing how bias may have entered the final product, but it’s there. How do you fix that? Again, that’s for future researchers.

Summoning up the major areas for AI, we see they include electronic health records, patient communication (EPIC is used by over 80 health systems to draft responses to questions), insurance and billing, scheduling and resource management, data, extraction, and analysis, and the assumed reducing burnout of staff.

What can AI do? What are the benefits to patients in terms of quality, efficiency, and accessibility of healthcare? Some of the key benefits include:

1. Improved diagnosis and treatment by analyzing medical data quickly with AI algorithms to detect patterns in medical imaging, x-rays, and MRIs.
2. Personalized medicine where treatment plans can be tailored to individual. Patient needs in terms of genetic information, lifestyle, and other health data. This should address decreasing side effects.
3. Predictive analytics where potential health issues are discovered through the analysis of trends and patterns found in patient data. This would then permit early intervention and preventive care, thereby improving outcomes.
4. Enhanced patient monitoring through AI-powered wearable devices and remote monitoring tools that continuously track a patient’s vital signs and other health metrics. These interventions are particularly helpful in cases of chronic conditions.
5. Streamlined administrative processes, such as appointment scheduling, billing, and record management, which reduce wait times and can improve patient experience.
6. Virtual health assistance in AI Chatbot provides instant access to medical information, answers to health-related questions, and guidance on managing conditions with self-care.
7. Mental health support through apps that provide cognitive, behavioral therapeutic interventions, mood tracking and stress management techniques. This would then make mental health care more accessible and immediate.
8. Drug discovery and development through AI greatly accelerates this process, leading to innovative treatments and medications more quickly.

Risks for AI in Healthcare

While the benefits of AI in healthcare are impressive and can be utilized in a range of areas, there is no such thing as a free lunch and we must always consider the risks that are attached to this type of healthcare management. What are we concerned about? Here are some of the things:

1. Bias and inequity where the system can perpetuate or even exacerbate biases in healthcare there why bringing about unequal treatment and outcomes for a different patient groups, specifically minorities and underserved populations.
2. Accuracy and reliability is of great concern because any system can make errors such as misdiagnosis or incorrect treatment recommendations. This area is especially crucial and must be carefully managed and monitored.
3. Data privacy and security must address and prevent breaches and misuse of data.
4. Transparency and explainability would be areas in which those who are not highly conversant with algorithms would experience difficulty understanding how conclusions were derived from the data.
5. Regulatory and legal challenges must be carefully addressed in new guidelines that address AI-specific issues.
6. Over-reliance on AI presents a risk leading to automation bias, where clinicians cannot see errors made by AI and assume it is infallible.
7. Integration with existing systems is especially important for any issue because not all algorithms can speak to other algorithms and require what is known as an API, which allows it to integrate one system into another, whereby they work well together.
8. Ethical concerns would be in informed consent, patient autonomy, and the potential for AI to replace human jobs.
9. Economic and accessibility issues revolve around the development and implementation of AI into healthcare solutions for smaller or resource-limited healthcare providers.

The age of automation and artificial intelligence is here to stay, and as more creative approaches are designed, using less energy and providing faster, more accurate results, we will see it spreading everywhere. As long as you have any type of computerized equipment, you will have access to AI-assisted learning and data analysis.

In fact, an algorithm can use your data to learn as it is trained, so that each of us can be a research subject without our knowledge. This may sound dystopian, but it does have major advantages because garnering large groups of people for research can be an extremely difficult, time-consuming, and expensive process. Each step we take that shortens any of these areas in terms of time or money means better health for all of us.

AI may not be the wonderful solution we seek currently, but it has extraordinary potential in the future as long as we balance its merits and potential challenges well.