My talk has two distinct halves. First, I will present work from my postdoc around the question of “optimality” in economic decision-making. I will show that two simple assumptions – a noisy decision process and the desire to predict outcomes accurately – can account for well-known biases in economic decisions. I will first show how these effects are predicted in simulations, before showing in two experiments that humans behave in ways consistent with the simulations.
In the second half of my talk I will address an entirely different, new strand of research. I have since left academia for a research position in the healthcare industry, where our team seeks to improve treatment and outcomes for patients with mental health problems. I will present some initial results around how information gathered via an online referral chatbot can be used with machine learning tools to assist clinicians in diagnosing and improving treatment outcomes.

 

Keno Jüchems
Machine Learning Scientist
Limbic Ltc, London

Keno Juchems – Optimal economic decision-making under noise and machine learning to improve outcomes in mental health