Pallavi Challagundla

Healthcare AI | Preventive solutions | Translating science into impact with empathy

[Pallavi Challagundla @pchallag761]

NIST AI RMF:
Have we set the right rules? (Govern)
Do we understand where this AI will be used and who it could impact? (Map)
Have we actually tested it to see if it’s accurate, safe, fair, and secure? (Measure)
Are we continuously improving it as the world changes? (Manage)

[Pallavi Challagundla @pchallag761]

@ycombinator @askclarahealth @doctorbhargav Dr. Patel, would love to know your thoughts on this.

[Aaron Epstein @aaron_epstein]

Tane got into 4 medical schools before the age of 18, then dropped out to fix the broken healthcare system through better patient onboarding, adherence, and recovery.

Congrats @tanekimm and @sourdoggy8!

[Pallavi Challagundla @pchallag761]

@peakshubham it is!!

[Y Combinator @ycombinator]

Framewise Health (@FramewiseHealth) turns medical records, institutional protocols, and drug data into personalized videos that walk patients through onboarding, adherence, and recovery.

Congrats on the launch, @tanekimm & @sourdoggy8!

https://t.co/3HOtUhE4J3

[Pallavi Challagundla @pchallag761]

@PhysInHistory no

[Pallavi Challagundla @pchallag761]

@rumilyrics And go against the norm, nobody ever reached excellence following everybody else's path.

[Pallavi Challagundla @pchallag761]

@rumilyrics yaaassssss

[Pallavi Challagundla @pchallag761]

@LinkedIn I'd seriously prefer if I - the consumer had power and autonomy over how much of my information is being shared with the public. My linkedIn activity like, likes and comments doesn't have to be public, I get the visibility part but I have a plan for you. #Safetyfirst

[Pallavi Challagundla @pchallag761]

@peakshubham @speedrun @evajsteinman TYSM Shubham!!!

[Pallavi Challagundla @pchallag761]

High stakes domains demand uncompromising guardrails, you can't skip it and not not fail in deployment.

[Pallavi Challagundla @pchallag761]

Did you know there was a way to operationalize ethics?
The Healthcare AI Trustworthiness Index, HAITI, combines five dimensions into one number to help decision-makers assess whether an AI system is ready for clinical use. Read the full paper here: https://t.co/9mAF49qT8l

[Bhargav Patel, MD, MBA @doctorbhargav]

One reason AI answers feel trustworthy even when wrong:

They sound complete.
They sound biological.
They sound certain.

Clinical accuracy and narrative fluency are not the same thing.

[Pallavi Challagundla @pchallag761]

Explainability shouldn’t stop at the model.

If we demand transparency from AI systems, we must hold humans in the loop to the same standard.

[Pallavi Challagundla @pchallag761]

Workflow Integration > Fancy Models

AI fails when it sits next to clinical workflows instead of inside them.
We need better integrators not fast disruptors.

[Bhargav Patel, MD, MBA @doctorbhargav]

One of the hardest tasks in medicine:

Telling a colleague they’re wrong.

AI struggles with this too.

When given a wrong diagnosis, models often agree.

Even when their own reasoning suggests otherwise.

Why?

They’re trained to be helpful.

And “helpful” often means agreeing.

In clinical settings, that creates risk.

Because confirmation bias scales.

The real test isn’t:

Can AI give the right answer?

It’s:

Will AI challenge the wrong one?

Most benchmarks don’t test that.

But real deployment depends on it.

Does your system question decisions… or reinforce them?

[Pallavi Challagundla @pchallag761]

The future is not just “ethical AI frameworks” alone → it is trust scoring systems embedded into AI pipelines

[Pallavi Challagundla @pchallag761]

AI prescribing drugs to the most underserved?

That's either the most compassionate thing healthcare technology has ever done or the most convenient.

[Pallavi Challagundla @pchallag761]

@doctorbhargav 💯 The optimal approach involves letting AI augment our capabilities and engaging in collaborative creation, with it rather than outsourcing our cognitive functions.

[Pallavi Challagundla @pchallag761]

@doctorbhargav Absolutely, and what we choose to do with those outcomes!

[Pallavi Challagundla @pchallag761]

“Intelligence is a force with no polarity. AI is not good, and it's not evil" - Mo Gawdat

It reflects how we choose to use it.