In my Padlet I’ll show three clear use cases:

• Diagnostic imaging: AI helps flag tumors or bleeding so radiologists can look sooner.

• Personalized medicine: Tools like Tempus use data from labs and records to match patients to targeted drugs or trials.

• https://www.tempus.com/

• Patient monitoring: Companies such as Medtronic use AI with sensors to track chronic conditions between visits, which can alert the care team before things get serious. 

AI isn’t perfect but it learns from data, so if the data are biased or messy the tool can be unfair or wrong. We need strong rules for safety privacy and equity, and the FDA keeps a running list of AI medical devices it has cleared. That helps hospitals know what’s trustworthy.

Overall, I’m interested in healthcare AI because it mixes science and empathy. If we build it with care and test it well, making it fair, and keeping humans in it. We can help more people get the right care at the right time.

Tech used: Deep learning trained on thousands of labeled images.

Benefits: Faster triage and fewer misses and more consistent reads. Real world studies show the workflow gains and accuracy support when used

Limits/risks: Needs high quality diverse data. False alarms can add work; final call should remain with clinicians. The FDA tracks cleared AI imaging tools to guide safe use.  

Image idea: CT scan with AI heat map overlay.

https://www.youtube.com/watch?v=76LqIY7uL2w

Tech used: Machine learning and large real world dataset some tools also use natural language processing to read notes.

Benefits: More tailored treatment options and quicker trial matching; programs like TEmpus +.

Limits/risks: Data privacy uneven access at smaller hospitals and bias if datasets don’t represent all groups.

Tech used: Time series ML and alert systems integrated with clinician dashboards.

Benefits: Earlier warnings, fewer ER visits, and personalized coaching for chronic diseases. Reviews highlight how continuous data can improve management.

Limits/risks: Battery connectivity, and false alerts equity concerns if devices don’t work as well across skin tones or body types. Companies such as Medtronic describe efforts to improve accuracy with AI. 

https://pmc.ncbi.nlm.nih.gov/articles/PMC11461032/

- generative AI assistants that draft notes after visit summaries, and discharge instructions to reduce burnout.  

- Multi modal models that combine images, text, labs, and sensor data to give a fuller picture of a patient.

- Predictive imaging which uses patterns in scans to forecast tumor type or treatment response, helping doctors plan sooner

- More FDA cleared tools and clearer pathways for updates as models learn over timee

https://www.nature.com/articles/s41746-025-01791-z

https://www.who.int/publications/i/item/9789240084759

Ethical points we should pay attention too.

• Bias and fairness- Some medical tools have shown accuracy gaps for darker skin, reminding us to build and test AI on diverse data. 

• Privacy and consent- Health data are sensitive so people must know how their data are used.

• Transparency and accountability- Patients deserve to know when AI is involved, and humans must stay responsible for final decisions.

• Safety and monitoring: Hospitals should track performance after deployment

My take is that AI should definetely help, but not replace, clinicians. If we follow principles and protecting human autonomy, promoting well being, being transparent and making systems fair. AI can earn trust and do real good.

https://www.mckinsey.com/mhi/our-insights/heartbeat-of-health-reimagining-the-healthcare-workforce-of-the-future

Access and equity- AI powered telehealth and remote monitoring can help rural and busy clinics reach patients sooner. But if tools work better on some groups than others gaps could grow. That’s why testing across diverse populations matters and why we need public reporting on performance.  

Safety and trust- FDA clearance and hospital oversight can lower risk but people should still be told when AI is used and how their data are protected. Clear consent should be used to build trust.

Costs: In the long run catching problems earlier and avoiding readmissions can save so much money. In the short run, buying and maintaining AI systems can be expensive, so hospitals should prove value with real outcomes.

My view- AI in healthcare is like a powerful tool kit. In the right hands, it speeds up care and makes it fairer. The goal is smart adoption to test it, explain it, and keep people at the center.

1. Wenderott, K., et al. (2024). Effects of AI implementation on efficiency in clinical imaging. npj Digital Medicine. https://doi.org/…

   Rationale: Gives real world evidence that AI can speed up imaging workflows. 

2. Najjar, R., et al. (2023). Redefining radiology: A review of AI in medical imaging. Diagnostics. https://pmc.ncbi.nlm.nih.gov/articles/PMC10487271/

   Rationale: Clear overview of how deep learning supports image analysis and where it falls short. 

3. World Health Organization. (2021). Ethics and governance of artificial intelligence for health. https://www.who.int/publications/i/item/9789240029200

   Rationale: Baseline ethics principles for safe and fair and transparent health AI. 

4. World Health Organization. (2025). Guidance on large multi-modal models in health. https://www.who.int/publications/i/item/9789240084759

   Rationale: Up-to-date guidance on next gen models that mix text images and more. 

5. U.S. Food and Drug Administration. (2025). AI/ML-Enabled Medical Devices. https://www.fda.gov/medical-devices/…

   Rationale: Confirms what AI medical devices are actually authorized in the U.S. 

6. Tempus. (2022–2025). AI-enabled precision medicine & Tempus+ program. https://www.tempus.com/

   Rationale: Explains how Tempus uses real world and genomic data to support care and research. 

7. Jafleh, E. A., et al. (2024). The role of wearable devices in chronic disease management. Sensors (via PMC). https://pmc.ncbi.nlm.nih.gov/articles/PMC11461032/

Rationale: Shows why continuous monitoring plus AI can improve longterm care.