CNN
–
Is my doctor using artificial intelligence to diagnose me during our appointment? Does he record our conversations and create an AI summary of our visits?
The use of AI healthcare is still new enough that many people may not know what to do with it. A 2023 survey found that most Americans expressed “significant discomfort” about the idea of doctors that can help them use AI to manage their care. However, there is little chance that AI will disappear. The use of AI applications in healthcare is increasing, and it is important for patients to understand the uses that can improve care and why they are ongoing attention.
I wanted to know if AI is already helping with diagnosis and directly assisting with treatment, and what clinicians think about AI use. And finally, what areas of concern are? And what is being done to address them?
To guide us with these questions, I spoke with Dr. Leana Wen, a CNN wellness expert. WEN is an emergency physician, an auxiliary associate professor at George Washington University and a senior non-resident fellow at the Brookings Institute, and her research includes crossroads of technology, medicine and public health. She was previously a health committee member in Baltimore.
CNN: What do patients think about the various uses of AI in healthcare?
Dr. Lean Wen: First, it helps you understand the difference between predictive and generative artificial intelligence, or AI.
Predictive AI uses mathematical models and pattern recognition to predict the future. For example, predictive AI algorithms can identify whether patients with pneumonia are most likely to require hospitalization.
Let's say you're a patient. Using past experience with many other patients with similar conditions such as pneumonia, diabetes, heart disease, etc., the algorithm may come up with a care plan for you based on factors that may affect your illness. This algorithm can help you determine, for example, whether you need to be hospitalized, which treatments are most likely to be effective in your particular situation.
Generation AI uses large-scale language models to generate human-like interactions. Many people may be familiar with ChatGpt. This is a form of generator AI that answers user questions in a conversational way. Genetic AI can summarise a large amount of information in a very short period of time and far outweigh that of humans. Several studies suggest that generative AI models can pass medical licensing exams and generate easy-to-understand and well-written patient instructions on a variety of topics.
However, there is concern that these models can “hastised,” misleading and inaccurate responses. Additionally, how well a model works both predictive and generative AI depends on the data trained. When assessing the utility of AI in healthcare, it is important to examine each AI tool individually to understand how it was developed and in what circumstances it will be used.
CNN: How is AI already used to diagnose patients?
Wen: There are several well-validated examples of predictive AI used to enhance and improve diagnosis.
Get colonoscopy, the gold standard for diagnosing colorectal cancer. During the procedure, the doctor passes through the colon and through a long tube to manually identify and remove cancerous or cancerous polyps.
AI can “train” to identify polyps and flag them during colonoscopy. Multiple randomized controlled studies conducted in health systems around the world have shown that increasing colonoscopy to use AI significantly reduces the MISS rate of potentially cancerous lesions.
Similarly, AI is used to help read mammograms, an important screening tool for breast cancer detection. Research has shown that AI-supported mammography screening is as accurate as having two trained radiologists read the study, and may even improve cancer detection while reducing clinician workload. The US Food and Drug Administration has already allowed around 20 AI products to help screen mammogram cancer, but adoption is limited due to the potential extra costs associated with deploying them in clinical practice.
CNN: How is AI used to direct treatment?
Wen: One use case is a predictive AI algorithm developed by researchers at Johns Hopkins University, which identifies hospitalized patients at high risk of developing sepsis. This early warning system has been deployed in multiple hospitals and has been found to reduce the time it takes to detect sepsis and start antibiotics and other treatments.
Kaiser Permanente also deploys predictive AI tools to look for signs of hospitalized patients' deterioration. If data refers to a worsening clinical condition based on patient vital signs, clinical tests, nurse reports, and other factors, the system issues an alert to ensure that the patient is evaluated promptly by an interpersonal clinician. This tool was associated with significantly lower mortality rates.
AI is shaping cancer treatments in the UAE
CNN: How do I know how my provider's office uses AI for diagnostics and other tasks? What can you do to protect your privacy?
Wen: Much of what we now call AI has been around for a while. For example, prediction algorithms have been used for many years to help coordinate treatment plans. The doctor's office is increasingly using AI to help draft email responses and to assist with documentation.
I think it makes sense to assume that your provider's office is using some form of AI for your care. You can ask your provider and also refer to the policies that the office asks to sign, which may contain permission to use certain technologies.
Your medical records are secured and protected by the Health Insurance Portability and Accountability Act, or HIPAA, which legally covers the protection of sensitive health information.
CNN: What do providers think about AI use? Does it reduce inefficiency and reduce the burden on documents?
Wen: I think clinicians were particularly pleased to have an AI application that cuts down documents and documents. Generator AI, for example, is used to support medical documents in a technology called Ambient AI. Essentially, this tool “listens” to conversations between doctors and patients and translates them into medical notes that doctors can edit. Research has shown that this surrounding AI scribe reduced note-taking time and was viewed positively by doctors and patients.
Also, doctors use the generated AI to assist with advance permits that they must send to their insurance companies to obtain permission for certain medications and treatments. This reduces the administrative burden of the clinic and perhaps even combats the burnout of healthcare workers.
CNN: Can AI be used by insurance companies to turn down claims?
Wen: That may be true. You can also imagine that the existence of AI is a major barrier to accessing insurance companies to discuss claims and other issues.
This is one of many concerns about using AI. Others include issues such as continuing to ensure privacy and data security, as well as the need to independently validate algorithms and transparently share these results. Technicians, clinicians and regulators recommend that you consider the many positive use of AI in healthcare while also rigorously studying each tool and deploying it carefully.
