Session 6/7 Assignment: AI in Industry and Society - Healthcare by Andrew Segovia

Personalized Overview.

andrewbaque2 — 2025-10-10 23:27:17 UTC

I decided to go ahead and choose the Healthcare industry because that is the one that is most relevant to me (as a premed student), the one that interests me the most regarding AI, and the one that I personally feel that will make the biggest (or one of the biggest) impacts in the future utilizing AI. I was already somewhat familiar with AI being used in healthcare (with things like Mammographies and detecting things like diseases early on), thus I thought it would make the most sense to choose this industry as I felt it would not only have a lot of scholarly articles, but many potential AI applications already. I think in general this was a good opportunity to research and learn more about AI in healthcare today (and its potential future applications) since it's clear that AI will not only make a profound impact and expand many other sectors of life (like the entertainment industry), but especially professional sectors like medicine and healthcare as a whole. My forthcoming padlet highlights three major and well-known applications of AI in healthcare that illustrate both promise (regarding the future of AI in this industry) and impact, such as AI-enabled mammography (which aids in breast cancer detection), and also the potential societal impacts that these applications may have on not only society but our future.

AI in Diagnostic Imaging (Breast Cancer Screening/Mammography)

andrewbaque2 — 2025-10-12 02:05:47 UTC

"AI in Diagnostic Imaging (Breast Cancer Screening/Mammography)" refers to the use of AI (particularly things like Deep Learning and ML, machine learning) to analyze mammography images (specialized X-ray pictures of the breast, done as part of a mammogram procedure for breast cancer screening and diagnosis). It is not a replacement for a human radiologist (person who specializes in interpreting the images of the breast in order to diagnose), but rather a support tool designed to enhance accuracy, efficiency, and patient care. Prospective evidence shows AI support can raise breast cancer detection in population screening without increasing recalls. In a study that had 24,543 women participants for example, radiologists using "AI-CAD" detected 140 cancers vs. 123 without AI (about 13.8% higher CDR or cancer detection rate) with no significant change in recall rate. This shows AI aid can significantly boost cancer detection while keeping false alarms low. The additional benefits that this provides, which can be influential to healthcare, are faster, more consistent reads and decreasing workload but a major concern some people have is the overreliance on AI and it potentially interfering with the jobs of radiologists in the future.

Sepsis Early-Warning Systems (Clinical Deterioration Prediction)

andrewbaque2 — 2025-10-12 02:06:10 UTC

Refers to automated tools powered by AI (through machine learning ML and Deep Learning) that continuously analyze a patient's electronic health record (EHR) data to predict when their condition is declining and to identify the early onset of sepsis (or other diseases/infections). A major study across seven hospitals showed that an AI early-warning score (eCART) actually beat the usual scoring rules when it came to predicting patient decline. This machine-learning model was much better at flagging which patients were at risk sooner (meaning higher "AUROC"), proving that AI can help clinicians catch potential problems earlier. The benefits of this is better accuracy and faster recognition but there are some challenges (such as the system needing to be fine-tuned to each hospital).

Ambient Clinical Documentation (AI Medical Scribes)

andrewbaque2 — 2025-10-12 02:06:27 UTC

Refers to AI (utilizing Large Language Models LLMs, Natural Language Processing NLP, and Automated Speech Recognition ASR) being used to listen in on the conversation between a clinician (like a doctor or nurse) and a patient, and then automatically generating the clinical note. This occurs silently and automatically in the background, without the need for the clinician to actively type or look at a computer screen. Ambient AI "scribes" are designed to tackle burnout by combining speech recognition and large language model (LLM) technology to automatically draft patient visit notes. For example, a JAMA study (across multiple centers) showed this greatly helps clinicians as physicians burnout dropped significantly (from 51.9% to 38.8% in a month), saving doctors about 0.9 hours a week on after-hours charting. The benefits are self explanatory: clinicians get more quality time with patients and spend less time on paperwork. But despite this, there still poses some considerable complications such as AI accuracy and privacy and consent (since the audio/what is being said is being recorded).

Future Trends and Ethical Considerations.

andrewbaque2 — 2025-10-12 02:06:48 UTC

I personally believe that the future trends and potential developments of AI in Healthcare can be massive. For example, AI scribes and clinical documentation using AI can be revolutionary for clinicians in terms of time efficiency and productivity (reducing documentation time and lower burnout rates could help hospitals and clinics as a whole). That being said, some of the ethical implications can already be noticed, such as patient privacy and transparency. The patient's conversation with the doctor would be recorded and analyzed/documented by the AI, thus the patients would need to be notified of this and there is also the risk of having a mistranslation/error with the AI getting something incorrectly, and the problems that can pose. Regardless, I do personally believe that AI scribes will be the future in some way and it will only continue to improve in accuracy. Another huge trend that has incredible and life-saving implications is the usage of AI in Diagnostic Imaging (mammography). We already have real-world data proving that AI support can help doctors find more cancers without increasing false alarms (and thus recalls). This changes healthcare as a whole, as now AI can be used to detect things like cancer at a better rate than most humans and this will only continue improving. With that being said, the implications that this can have on not only liability, but over reliance and career choices cannot be ignored. Overreliance on this technology could blur the lines in ethics, as if we are solely using the AI to detect cancer, what would that mean for radiologists and even complications (who would be held liable if the AI malfunctions and incorrectly detects or misses breast cancer for example). Although these two technologies previously discussed can be revolutionary for healthcare, and some may even argue that the pros outweigh the cons, we need to be careful on how they are utilized. We shouldn't want to build an overdependence on them, and we should also look to the ethics of utilizing them. Continuously monitoring drifts in accuracy and noting any disparities, as well as disclosing any AI involvement to clinicians and patients alike, is a good way to start in regards to ethics.

Societal Impact.

andrewbaque2 — 2025-10-12 02:07:01 UTC

The societal impacts that AI in general can have in healthcare is both intriguing but alarming. On one hand, technologies such as the "AI Medical Scribes" can reduce workloads and burnouts leading to greater patient care, productivity and efficiency. On the other hand, if we build an overreliance on technologies like the "Sepsis Early-Warning Systems (Clinical Deterioration Prediction)", we run the risk of the AI being wrong (or malfunctioning) and incorrectly raising unnecessary alarms (or missing problems later on) which will cause the public to lose trust. We also trend towards making some certain jobs obsolete and no longer needed, like radiologists for breast cancer detection. Even technologies like the AI medical scribes pose a risk towards patient (and even clinician) confidentiality and privacy. With that being said, the benefits that AI can have in not only research and patient care, but also patient convenience is astounding. For example, those with Parkinson's Disease (movement disorder of the nervous system that worsens over time) may be able to monitor progression on their phone (using machine learning to build a system). This could be life changing for patient care and convenience, since the patient themselves could directly monitor their treatment and progression at home along with physicians. I personally believe that with "guardrails" and the proper use of ethics (like notifying patients when AI is used), AI revolutionizes the future of healthcare and makes productivity and ease of access that much easier for both patients and physicians.

Reflection.

andrewbaque2 — 2025-10-12 02:07:15 UTC

I was already somewhat familiar with AI in healthcare prior to my research. I had heard/read about AI mammography and it being used to detect cancer and early diseases, as well as other uses like AI scribes, but I wasn't aware of how much impact it could all have surrounding healthcare and what implications it has for other similar areas in healthcare. For example, I never considered the impacts having ambient AI scribes could have on workload and efficiency, where burnout was reduced immediately upon utilizing this technology. I did learn a lot more about certain technologies like sepsis prediction, and how just having a high prediction score (AUROC) isn't the final goal but things like calibration (to hospital), alert fatigue, and local validation are as well. I did not encounter many challenges throughout this assignment, the only one being trying to somewhat understand the language used regarding some of this medical field related topics. I already do think AI has made a significant impact in the medical field in its short time being around, and I know it will only increase exponentially as time goes on. Other areas of the medical field that I did not touch on much like research are also being profoundly impacted by AI, which could only mean positive things for the future of patient care. I've always had a positive view of AI for the most part, especially since I already knew prior to the assignment that AI has been used to detect things like cancer, and this research has only solidified my reasoning that AI is the way forward for healthcare but with much needed guardrails to ensure ethics is maintained for patients and clinicians alike.

References.

andrewbaque2 — 2025-10-12 02:08:07 UTC

Chang, Y.-W., Ryu, J. K., An, J. K., Choi, N., Park, Y. M., Ko, K. H., … Han, K. (2025). Artificial intelligence for breast cancer screening in mammography (AI-STREAM): Preliminary analysis of a prospective multicenter cohort study. Nature Communications, 16, 2248. https://doi.org/10.1038/s41467-025-57469-3 [Rationale: Good source that gives real world evidence and studies on AI increasing cancer detection]

Radiological Society of North America. (2024, June 4). AI detects more breast cancers with fewer false positives. https://www.rsna.org/news/2024/june/ai-detects-more-breast-cancers [Rationale: Good source that talks about fewer false positives regarding AI in breast cancer detection]

Edelson, D. P., Churpek, M. M., Carey, K. A., et al. (2024). Early Warning Scores With and Without Artificial Intelligence. JAMA Network Open, 7(10), e2438986. https://doi.org/10.1001/jamanetworkopen.2024.38986 [Rationale: source showing ML scores outperforming "traditional" tools.]

Wixon-Genack, J., Wright, S. W., Cobb Ortega, N. L., Hantrakun, V., Rudd, K. E., Teparrukkul, P., Limmathurotsakul, D., & West, T. E. (2024). Prognostic accuracy of screening tools for clinical deterioration in adults with suspected sepsis in Northeastern Thailand: A cohort validation study. Open Forum Infectious Diseases, 11(5), ofae245. https://doi.org/10.1093/ofid/ofae245 [Rationale: source that discusses clinical deterioration prediction outside of the United States in East Asia.]

Olson, K. D., Meeker, D., Troup, M., et al. (2025). Use of ambient AI scribes to reduce administrative burden and professional burnout. JAMA Network Open, 8(10), e2534976. https://doi.org/10.1001/jamanetworkopen.2025.34976 [Rationale: source gives good numbers for a study showing reduced burnout and hours, good quantitative example.]

Tierney, A. A., Gayre, G., Hoberman, B., Mattern, B., Ballesca, M., Kipnis, P., Liu, V., & Lee, K. (2024). Ambient artificial intelligence scribes to alleviate the burden of clinical documentation. NEJM Catalyst Innovations in Care Delivery, 5(3). https://doi.org/10.1056/CAT.23.0404 [Rationale: credible source discussing early outcomes of ambient scribes and how it alleviates burden on clinicians.]

López González, L. (2024, December 10). 4 ways artificial intelligence is poised to transform medicine. UC San Francisco. https://www.ucsf.edu/news/2024/12/429031/4-ways-artificial-intelligence-poised-transform-medicine [Rationale: source gives good examples of AI transforming healthcare, especially for things like monitoring Parkinson's disease on your phone]

Overview Video.

andrewbaque2 — 2025-10-13 02:43:53 UTC