Applied machine learning scientist in health research

Health data is scattered.
I work on what we can still learn when it has to stay that way.

I'm Ashkan. I work on machine learning for health studies whose data sit in many places, including a lot of the work that has to happen before any model can run.

Now I am testing whether models that adapt to each site hold up in drug safety research, through an Innovation Mandate: a Flemish government grant, with Johnson & Johnson Innovative Medicine as industry partner.

Explore my work CV

Ashkan Pirmani, in an OHDSI Europe shirt and a Hasselt University lanyard, looking to the left of the camera.

Selected work

Four projects

All work

COVID-19 and multiple sclerosis

Four registries could not send their data, so I sent the analysis to them instead.

Registrymay shareRegistrymay shareRegistrymay not shareits records stay hereThe studyone data set7,757records representedRegistrymay shareRegistrymay shareRegistrymay not shareits records stay hereThe studyone data set7,757records represented

This registry is not allowed to share its records, so its patients are missing. Press Send the analysis to see what changes.

A simplified drawing. The counts are from the paper.

I designed and deployed this route. It added 3,527 records, 45.5% more than the study had without it. Most of the work was agreeing on definitions with eighteen registries, and running the analysis inside those four.

Informed MS International Federation advice · JMIR Medical Informatics, 2023

A project I stopped

I had built FL4E, and it worked. Then I started my next study, and I did not use it.

It was meant to save research teams from rebuilding the same foundations for every study across centers. But it was built for clicking, and my next study needed scripts. The data agreement also decided where my computing could run.

So I did not build a second version. I retired it, and wrote a paper on why a working tool still went unused.

Read the note And the time a simpler method was enough

Ashkan Pirmani at a lectern in front of a large screen. The slide reads: Good for all. Not always good enough for one.
Genova, Medical Informatics Europe 2026.
Five people crowded around a laptop at Hack4Health in Flanders Expo, Ghent. Ashkan Pirmani is seated at the left, talking with the others.
Hack4Health, Flanders Expo, Ghent.

About me

I thought my PhD would start with a USB stick.

Before health, I worked on Iran Post's parcel network, deciding where the sorting hubs should go. So when I started my PhD, I expected someone to give me the data, and I would get on with the modeling. That did not happen. For almost two years, most of my work was with clinicians, engineers, the people who look after the records, and their lawyers: who held the data, what they were allowed to do with it, and whether a word meant the same thing in two registries.

Read my story

Let's talk.

If your question depends on health data that can't simply be put in one place, I probably want to hear about it.

Email me

Opens your email app with the question in it. Nothing is sent from here.

TurkishMerhabaAzeri, my mother tongue, as written in Urmia: xoş gəldin, welcomeخوش گلدینAzerbaijani, as written in AzerbaijanSalamPersian, the language of school: dorudدرودEnglishHelloDutch, as said in FlandersDagyou
A greeting in each language I speak, where it lives.