Personalized Therapy for Opioid Use Disorder

1569 0

There is not a week goes by these days without an article about Opioids.

George Orwell’s observation (from many years ago) hit home:

We have developed a sort of compunction which our grandparents did not have, an awareness of the enormous injustice and misery of the world, and guilt-stricken feeling that one ought to be doing something about it, which makes a purely aesthetic attitude towards life impossible. No one, now, could devote himself to literature as singly-minded as Joyce or Henry James.

Can we ignore the epidemic of Opioid Use Disorder and simply go about living our lives?

Is there a moral responsibility to apply one’s skills and abilities to just help others and lessen their misery?

Even if you had nothing to do with what created that misery? That is, as Deborah Eisenberg’s book title: Your Duck is my Duck.

I remember reading Bill Clinton’s book Giving many years ago where he provided good examples of how one can use their time, skills and network, in addition to money, to help others.

What is the right amount of altruistic effort that one is obligated to expend as a fellow member of society?

Isn’t saving hundreds of lives through increased supply of organ and tissues for transplants enough?

Have I not built up sufficient inventory of कर्म points? 🤷🏽‍♂️

No, said my conscience: Think! You ought to contribute more to things that matter. 


Can I really have an amply lived life if I did not continue to use my imagination, skills, money and network to lessen the misery of others?

So, I read Beth Macy’s Dopesick and Lloyd Sederer’s The Addiction Solution.

And I started thinking: what can I do that is elegant and effective in combating these opioids, known among the families of the users as Oxycoffins?

And there it was!

Use wearables to predict the craving state of the individual, via machine learning and AI, and then intervene just as they are about to do something they have no self-control over.

Personalized Treatment for Opioid Use Disorder!

To me, this sounded like pre-cogs dealing with pre-crimes in the movie Minority Report.

Loved the idea.

This immediately brings up interesting questions.

What is the value of this timely information over current protocols of urine tests and self-reporting?

There are wearables with different accuracy levels and costs, and so which is the optimal one to use in presence of budget constraints?

How would one go about studying this in a systematic way with real-world data and field validation?

This application is a poster child for the use of Partially Observed Markov Decision Processes (POMDP) with budget constraints!

Kyra Gan, a PhD student, is building the MDP and POMDP models as part of her first year paper requirement.

Alan Scheller-Wolf, an expert in MDP and POMDP, has graciously agreed to co-advise Kyra with me.

We are off to the races!

Leave a Reply