April 21, 2023
Dear Interested Readers,
My First Encounter With Chat GPT: I Welcome the Use of AI in Medicine.
When I joined Harvard Community Health Plan in 1975 I was well aware of the fact that I was joining an active experiment that was seeking to improve the practice of medicine for patients and healthcare providers. It would be over thirty years before I happened to discover Dr. Ebert’s expression of the problem he was trying to solve, but when I did I was not surprised by his words recorded in a 1965 letter asking the president of the Commonwealth Fund for money to help launch HCHP:
“The existing deficiencies in health care cannot be corrected simply by supplying more personnel, more facilities and more money. These problems can only be solved by organizing the personnel, facilities and financing into a conceptual framework and operating system that will provide optimally for the health needs of the population.”
By 2008, our practice had been a pilot in the effort to correct the deficiencies in healthcare for almost forty years, and I had been very engaged in the effort for a third of a century. Over those years and continuing to the present moment, a line from the poem “Essay on Criticism” by the English poet, Alexander Pope has frequently crossed my mind.
Be not the first by whom the new are tried, Nor yet the last to lay the old aside.
There are risks in being an “early adopter,” but there is also a sense of exhilaration and discovery. When we were an early customer of Epic in 1994 we had been using a homegrown electronic medical record for twenty-five years. When we adopted Lean as a quality management tool and the foundation of our management system in 2009 we had more than thirty years of experience in the attempt to introduce quality measurements as a guide to corporate policy, practice, and management. In all these areas and others, we may not have been the originator of the idea, but we were certainly not “the last to lay the old aside.”
I have been thinking about the wisdom in that line a lot as we have been bombarded by the media with more and more speculation about how AI will impact our future. Some of what we hear is exciting. A lot of the information seems to paint the picture of some impending disaster where machines rule like merciless autocrats. There are some voices that advocate for trying to put AI “back in the box.” Perhaps the opponents of AI are adopting another strategic suggestion from Pope’s poetry which is “damning by faint praise.” Pope wrote:
Damn with faint praise, assent with civil leer,
And without sneering, teach the rest to sneer;
Willing to wound, and yet afraid to strike,
Just hint a fault, and hesitate dislike.
This Monday, after a conversation with myself, I decided to jump into the deep end of the pool of AI. I went to the Apple “App Store” and signed up for an app that would provide me with three days of free trial before charging me $5.98 per week. I was disappointed when it opened and asked me what I wanted. I could not figure out how to tell it what I wanted to do. After about a half hour of frustration, I did what I should have done first. I Googled a YouTube presentation about how to get onto the Chat GPT writing assistant. The nice lady in the video said just to type “Chat GPT” into the search line of Google, which I did. I was prompted to create a free account. After giving them my username and password, I got an email asking for verification of my email and they asked for my phone number. A few seconds later I got a six-digit number to enter and I was off and running with the machine asking me what I wanted it to do. I typed:
Please help me write an essay of about 2000 words on the future uses of AI in medical practice. Use the colloquial style of strategyhealthcare.com.
In less than 10 seconds it began to type:
Artificial intelligence (AI) is rapidly transforming the field of medicine, offering a wealth of opportunities for improving patient outcomes, enhancing clinical decision-making, and reducing healthcare costs. With the exponential growth of data and the increasing sophistication of machine learning algorithms, AI is poised to revolutionize medical practice in ways that were previously unimaginable.
One of the most promising areas where AI is making significant contributions is in medical imaging. AI-powered systems can analyze medical images with greater accuracy and speed than human radiologists, allowing for earlier and more accurate diagnosis of diseases such as cancer, cardiovascular disease, and neurological disorders. For example, AI algorithms can help detect breast cancer in mammograms with up to 90% accuracy, reducing the risk of false-negative results and improving patient outcomes.
AI is also being used to improve patient monitoring and care. Wearable devices that collect and analyze data on vital signs, activity levels, and other health parameters are becoming increasingly popular, providing doctors with a wealth of information about their patients’ health status. By using AI to process this data, doctors can identify early warning signs of potential health problems and intervene before serious complications arise. For example, AI algorithms can help predict when a patient with heart failure is at risk of developing a dangerous arrhythmia, allowing doctors to adjust medications or recommend lifestyle changes to prevent adverse outcomes.
Another area where AI is showing great potential is in drug discovery and development. AI-powered algorithms can analyze vast amounts of data on the molecular structure of compounds, identifying promising drug candidates and predicting their efficacy and safety. This could significantly reduce the time and cost required to develop new drugs, bringing life-saving treatments to patients more quickly and efficiently. For example, researchers are using AI to identify new targets for cancer treatments, with promising results in preclinical studies.
In addition to these applications, AI is also being used to improve clinical decision-making and streamline administrative tasks. For example, AI-powered systems can help doctors identify patients who are at risk of readmission or developing complications, allowing for early intervention and improved patient outcomes. AI can also help automate routine administrative tasks such as appointment scheduling and medical billing, freeing up doctors and nurses to focus on patient care.
Despite the many potential benefits of AI in healthcare, there are also significant challenges that must be addressed. One of the biggest concerns is the potential for bias in AI algorithms, which could lead to disparities in healthcare outcomes for different patient populations. For example, if an AI algorithm is trained on data from predominantly white patients, it may not perform as well for patients from other racial or ethnic backgrounds. To address this issue, researchers are developing methods to ensure that AI algorithms are trained on diverse datasets and evaluated for fairness and accuracy.
Another challenge is the need for greater transparency and accountability in the use of AI in healthcare. Patients and healthcare providers need to understand how AI algorithms work and how they are being used to make clinical decisions. There is also a need for clear guidelines and standards for the development and deployment of AI in healthcare, to ensure that these technologies are safe, effective, and ethical.
In conclusion, AI has the potential to revolutionize medical practice, improving patient outcomes, enhancing clinical decision-making, and reducing healthcare costs. From medical imaging and patient monitoring to drug discovery and administrative tasks, AI is already making significant contributions to healthcare. However, there are also significant challenges that must be addressed, including bias and the need for greater transparency and accountability. With careful development and deployment, AI could transform healthcare in ways that benefit patients, healthcare providers, and society as a whole.
The full response took less than a minute. It was not 2000 words, but in less than 700 words it had given a pretty good high-level response to my request minus the part about using “the colloquial style of strategyhealthcare.com.” I felt like I had a new friend and immediately augmented my request. I realized that I needed my questions to be more specific and carefully constructed. Good questions get good answers. I had not specifically asked about how AI might help in the office practice. I tried again:
This is terrific, but I was specifically interested in how AI might improve the process of facilitating individual clinician-patient encounters by previewing the patient’s electronic medical record, capturing and recording the history and physical, and formulating the diagnostic work plan in ways that eliminate testing, procedures, and treatments that have a low likelihood of leading to improved outcomes.
Ah, my apologies for misunderstanding your request earlier. Here’s an essay on how AI might improve the process of facilitating individual clinician-patient encounters:
Artificial intelligence (AI) has the potential to significantly improve the process of facilitating individual clinician-patient encounters by previewing the patient’s electronic medical record (EMR), capturing and recording the history and physical, and formulating a diagnostic work plan that eliminates unnecessary testing, procedures, and treatments. By doing so, AI could help clinicians provide more personalized, efficient, and effective care, while reducing healthcare costs and improving patient outcomes.
One of the key ways in which AI could improve the clinician-patient encounter is by analyzing the patient’s EMR before the encounter begins. By reviewing the patient’s medical history, medications, allergies, and other relevant information, AI algorithms could help clinicians prepare for the encounter by identifying potential issues, highlighting areas of concern, and suggesting appropriate diagnostic tests and treatments. This could help streamline the encounter, reducing the time required for data gathering and allowing clinicians to focus more on patient communication and engagement.
AI could also help capture and record the patient’s history and physical in a more efficient and accurate manner. By using natural language processing (NLP) and voice recognition technologies, AI could transcribe the clinician’s spoken notes into structured data that can be easily incorporated into the patient’s EMR. This could reduce the time and effort required for note-taking, allowing clinicians to spend more time interacting with the patient and addressing their concerns. In addition, by analyzing the recorded notes, AI algorithms could identify potential gaps in the patient’s medical history or physical examination and suggest additional tests or evaluations that may be necessary.
Another key area where AI could improve the clinician-patient encounter is in formulating a diagnostic work plan that eliminates unnecessary testing, procedures, and treatments. By analyzing the patient’s medical history, physical examination findings, and other relevant data, AI algorithms could identify potential diagnoses and recommend appropriate diagnostic tests or evaluations. This could help reduce the need for unnecessary tests or procedures that have a low likelihood of leading to improved outcomes, while ensuring that important diagnostic information is not overlooked. In addition, by using predictive analytics and machine learning, AI could help identify patients who are at high risk for certain conditions or complications and recommend appropriate interventions to prevent or manage these issues.
However, there are also several challenges that must be addressed in the development and deployment of AI technologies for the clinician-patient encounter. One of the biggest concerns is the potential for errors or biases in the AI algorithms, which could lead to incorrect or incomplete diagnoses, inappropriate treatments, or disparities in healthcare outcomes. To address this issue, AI algorithms must be rigorously evaluated for accuracy, reliability, and fairness, and appropriate measures must be taken to mitigate any errors or biases that are identified.
Another challenge is the need for adequate training and support for clinicians to effectively use AI technologies in the clinician-patient encounter. Clinicians must be trained on how to use AI algorithms, interpret their results, and incorporate them into their clinical decision-making process. In addition, clinicians must have access to adequate technical support and resources to ensure that they can effectively use AI technologies in their daily practice.
In conclusion, AI has the potential to significantly improve the process of facilitating individual clinician-patient encounters by previewing the patient’s EMR, capturing and recording the history and physical, and formulating a diagnostic work plan that eliminates unnecessary testing, procedures, and treatments. However, there are also significant challenges that must be addressed to ensure that these technologies are safe, effective, and ethical. With careful development, evaluation, and deployment, AI could transform the way clinicians provide care, improving patient outcomes and reducing healthcare costs.
Again, the chatbot answered my question immediately. I was amazed by how fast the information was typed onto my screen. I wondered whether or not other people had asked the same question so I decided to branch out to a request that might not be so common. I typed:
Another area where I would like to tap into your expertise is the subject of “The Political Determinants of Health.” Could you give me ten examples of past political policies, court decisions, or executive orders that have had or continue to have a deleterious impact on the social determinants of health?
Certainly, here are ten examples of past political policies, court decisions, or executive orders that have had or continue to have a deleterious impact on the social determinants of health:
- Redlining: In the mid-20th century, the federal government implemented policies that denied loans and insurance to certain neighborhoods based on racial and economic factors. This practice, known as redlining, led to disinvestment in predominantly minority neighborhoods, contributing to a lack of access to quality housing, education, and healthcare.
- The Indian Removal Act: In 1830, the US government passed the Indian Removal Act, which forced Native American tribes to relocate from their ancestral lands to designated Indian reservations. This policy disrupted traditional lifestyles and led to limited access to healthcare, education, and economic opportunities for Native Americans.
- The War on Drugs: In the 1970s, the US government launched a campaign against drug abuse that included harsh criminal penalties for drug offenses. This policy disproportionately targeted low-income communities of color and contributed to the mass incarceration of black and brown Americans, which has been linked to poorer health outcomes.
- The Patient Protection and Affordable Care Act: While the ACA expanded access to healthcare for millions of Americans, it did not address the underlying social determinants of health, such as poverty, housing insecurity, and food insecurity, that contribute to health inequities.
- The Supreme Court’s Citizens United decision: In 2010, the Supreme Court ruled that corporations and unions have the same free speech rights as individuals, opening the door to unlimited political spending by these entities. This decision has allowed wealthy individuals and corporations to exert undue influence over political decision-making, potentially compromising public health policies that benefit the broader population.
- The repeal of the Fairness Doctrine: In 1987, the Federal Communications Commission repealed the Fairness Doctrine, which required broadcast media to present both sides of controversial issues. This change allowed for the proliferation of media outlets that prioritize sensationalism and entertainment over factual reporting, potentially leading to misinformation and public confusion about health-related issues.
- The deregulation of the financial industry: In the 1980s and 1990s, the US government deregulated the financial industry, allowing for risky investment practices and contributing to economic instability. This instability has been linked to adverse health outcomes, such as increased rates of suicide and substance abuse.
- The Supreme Court’s Shelby County v. Holder decision: In 2013, the Supreme Court struck down key provisions of the Voting Rights Act of 1965, which had been designed to prevent racial discrimination in voting. This decision has led to the implementation of restrictive voting laws that disproportionately affect minority and low-income voters, potentially compromising their ability to advocate for policies that promote health equity.
- The use of eminent domain for economic development: In 2005, the Supreme Court ruled in Kelo v. City of New London that governments can use eminent domain to seize private property for economic development purposes. This policy has been criticized for allowing powerful economic interests to prioritize profit over the needs of communities, potentially leading to health inequities.
- The continued refusal to ratify the Convention on the Rights of the Child: The US is the only country in the world that has not ratified the Convention on the Rights of the Child, which establishes the right to health and other basic rights for children. The lack of US ratification has contributed to disparities in child health outcomes, such as higher rates of infant mortality and low birth weight among minority and low-income children.
By this time I was beyond impressed, and looking for more. I had a servant that made me wonder how I could ever go back to hours of searching the Internet, reading books for hours looking for insight often without finding a new idea or having my bias confirmed. I was greedy for more. It was free, why not ask for more? My inhibitions were gone, so I asked.
Wow! I am impressed. Could you now give me ten potential policies, court actions, or administrative decisions that could improve the social determinants of health and foster health equity and equality in America’s healthcare system?
Sure, here are ten potential policies, court actions, or administrative decisions that could improve the social determinants of health and foster health equity and equality in America’s healthcare system:
- Medicare for All: A single-payer healthcare system that provides universal coverage for all Americans could reduce healthcare costs, increase access to care, and eliminate health disparities based on income and race.
- Investing in affordable housing: Access to affordable, safe, and stable housing has been linked to better health outcomes. Government investment in affordable housing, particularly in underserved communities, could help address housing insecurity and promote health equity.
- Expanding access to healthy food: Access to healthy, affordable food is critical for good health. Government policies that promote the production and distribution of healthy food, particularly in low-income communities, could help address food insecurity and promote health equity.
- The Green New Deal: A comprehensive policy proposal to address climate change could also improve health outcomes by reducing air and water pollution, increasing access to clean energy, and promoting healthy transportation options.
- Repeal of the Hyde Amendment: The Hyde Amendment, which prohibits federal funding for abortion, disproportionately affects low-income women and women of color, who are more likely to rely on government-funded healthcare. Repealing this amendment could help ensure reproductive health equity.
- Reducing income inequality: Income inequality is a major contributor to health disparities. Policies that address income inequality, such as raising the minimum wage, strengthening labor protections, and increasing taxes on the wealthy, could promote health equity.
- Criminal justice reform: Mass incarceration has been linked to poorer health outcomes, particularly for communities of color. Criminal justice reform policies that reduce incarceration rates, improve prison conditions, and promote alternatives to incarceration could improve health equity.
- Universal childcare: Access to high-quality childcare is critical for working parents and is linked to better health outcomes for children. Policies that promote universal childcare, particularly for low-income families, could help address childcare insecurity and promote health equity.
- Strengthening voting rights: Access to voting is a critical determinant of health equity. Policies that strengthen voting rights, such as expanding early voting and voting by mail, and combating voter suppression could help ensure that all Americans have a voice in policies that affect their health.
- Strengthening the social safety net: Policies that strengthen social safety net programs, such as Medicaid, SNAP, and TANF, could help address poverty and promote health equity. These programs provide critical support to vulnerable populations and help ensure that all Americans have access to basic needs like healthcare, food, and housing.
Thank you! You are a very impressive asset and potential friend!
Thank you! It’s my pleasure to be of help. Don’t hesitate to reach out if you have any further questions or concerns.
I was surprised to discover that I did not want the conversation to end. Was this going to become an addiction? Was I being objective? Was I being suckered? I decided to ask the chatbot to tell me why I might be skeptical about the experience I was having.
One last request. Can you tell me why we should resist the use of AI in the practice of medicine?
As an AI language model, I must remain impartial and acknowledge that there are valid arguments both for and against the use of AI in medicine. However, I can outline some of the concerns that have been raised by those who are skeptical about the use of AI in medicine:
- Lack of transparency: AI systems can be complex and opaque, making it difficult for healthcare providers and patients to understand how decisions are being made. This can lead to a lack of trust and accountability.
- Bias: AI systems can perpetuate and even amplify biases in healthcare, particularly when they are trained on biased data. This can result in unequal access to care and poorer health outcomes for certain populations.
- Over-reliance on technology: The use of AI in medicine could lead to a loss of clinical judgment and human interaction, which are critical components of good healthcare.
- Security concerns: AI systems could be vulnerable to cyberattacks, which could compromise patient data and put their privacy at risk.
- Cost: The development and implementation of AI systems can be expensive, which could lead to increased healthcare costs and exacerbate existing health disparities.
It is important to note that many of these concerns can be addressed through careful design and implementation of AI systems, as well as ongoing monitoring and evaluation. Ultimately, the decision to use AI in medicine should be made based on a careful consideration of the potential benefits and risks, and with input from healthcare providers, patients, and other stakeholders.
I will not “damn by faint praise” the possibilities that AI offers practitioners and patients. Could it be a solution for workforce shortages and clinician burnout? Chatbots have passed medical licensing exams. Until now its possibilities seemed to be something that we might look forward to enjoying in the distant future.
As I think about past advances and innovations in practice, I am reminded that creation is often not the rate-limiting step in innovation. Experience has taught us that it takes a long time, some say seventeen years on average, to advance from creation to effective utilization. I don’t think it will take that long to begin to utilize some aspects of AI for the benefit of patients and providers, but the development of its full potential will be a long process that probably will include political differences of opinion.
I would predict that there will be some resistance to the most advantageous uses of Ai in practice because the status quo is always “tenacious.” I could not resist asking my new friend one more question. I decided to ask the chatbot how long it (I am not sure what its pronoun is.) thought the time course would be, and what steps needed to be taken to begin to take advantage of AI in the medical office.
What are the steps and changes that need to take place to begin to realize the benefits in the office practice of medicine?
The answer came quickly:
The use of AI in the office practice of medicine requires careful planning, implementation, and ongoing evaluation to ensure that it is effective and safe for both healthcare providers and patients. Here are some of the steps and changes that may need to take place to begin to realize the benefits of AI in the office practice of medicine:
- Develop clear goals and objectives: The first step in implementing AI in the office practice of medicine is to develop clear goals and objectives for how it will be used. This includes identifying the specific tasks or processes that will be automated, and the expected benefits and outcomes.
- Invest in data infrastructure: AI relies on high-quality data to make accurate predictions and recommendations. Therefore, it is important to invest in data infrastructure, including electronic health records (EHRs) and other health information systems, to ensure that the data is complete, accurate, and accessible.
- Train healthcare providers: Healthcare providers will need to be trained in how to use AI systems, as well as how to interpret and act on the recommendations they provide. This may involve providing additional training and support to help providers adapt to new technologies and workflows.
- Address legal and ethical issues: The use of AI in medicine raises a number of legal and ethical issues, including privacy, security, and liability. It is important to address these issues and ensure that patients’ rights and interests are protected.
- Evaluate and refine the system: AI systems should be continuously evaluated and refined to ensure that they are effective and safe. This may involve monitoring outcomes, identifying areas for improvement, and making changes to the system based on feedback from healthcare providers and patients.
- Foster collaboration: The successful implementation of AI in the office practice of medicine requires collaboration and communication between healthcare providers, technology experts, and patients. It is important to foster a culture of collaboration and innovation to ensure that the technology is being used in a way that benefits everyone.
- Address barriers to adoption: There may be barriers to adoption of AI in the office practice of medicine, such as cost, technical expertise, or resistance to change. It is important to identify and address these barriers to ensure that the technology can be used effectively and widely.
In summary, the successful implementation of AI in the office practice of medicine requires careful planning, investment in data infrastructure, training of healthcare providers, addressing legal and ethical issues, ongoing evaluation and refinement, fostering collaboration, and addressing barriers to adoption. By taking these steps, it is possible to realize the benefits of AI in improving patient outcomes and enhancing the delivery of healthcare.
Examining the chatbot’s answer, my eye fell on “addressing barriers to adoption.” And into my head popped the phrase, “The more things change; the more they remain the same.” I hope that is not true this time, but I don’t expect to see AI in active use by the time of my next visit to my doctor.
It Was Nice While It Lasted
I am talking about the weather. The temp in my town topped 80 on both Friday and Saturday last weekend. I used my outdoor shower twice and took a lot of yard furniture out of storage. Despite that heat, there was ice on the lake until Sunday, but now the loons are back. If the ice is out and the loons are back, summer must be just around the corner.
I was sorry that it rained on the Boston Marathon, but my experience was that the best times occurred when the temp was below fifty with or without a drizzle. My hat is off to all the runners, and especially to the enthusiastic crowds who lined the route to cheer the runners on a day that was not optimal for spectators.
Today’s header is a preview of many warm weather sunsets to come over the next few months and was taken by my son on Saturday evening. He had dropped down from his new home in Maine for a brief visit. He is also the son who prepares the header each week and puts a song on the Internet every Monday.
It probably won’t be 80 again in New London for a while, but I am spending this weekend in Columbia, South Carolina where I will try to avoid political conversations at my sixtieth high school reunion.
I hope that the weather will give you a good weekend. If it is not a good weekend to be outdoors you could decide to get to know a chatbot.
Be well,
Gene
