Predicting medication response in ADHD through computational modeling of the Continuous Performance Test

ADHD is associated with negative consequences both for those diagnosed and the society in general. Medicines can reduce symptoms, improve daily functioning and increase quality of life. However, it is estimated that about one in three patients with ADHD do not respond satisfactorily to ADHD-medication. Early detection of non-responders could significantly improve their treatment, but finding early indicators of medication response has proven difficult. The current project takes a novel approach to prediction of medication response. By using computational modeling of decision making we will analyse already collected data from 250 adult ADHD patients. Patients in this study performed a computer-based test, the Continuous Performance Test (CPT), both before and 6 weeks after starting medication treatment. Preliminary results reveal that by analysing data with computational model, in contrast to standard analyses, responders and non-responders can be separated based on performance measures. Adapting the analysis tool and further investigating the collected data will be performed under the supervision of Professor Michael Frank at Brown University. Dr. Frank is one of the most prominent researchers in Computational Psychiatry, a recent field in which computational models of cognition and brain function are used to better understand psychiatric disorders. Another goal for the project is to make the analysis tool available for other researchers to analyse CPT-data. The next phase of the projects involves using the modified model to analyse data from other clinical groups. This work will be performed through close collaboration with researchers at the Department of Psychology at the University of Oslo and at Oslo University Hospital. The goal of this phase is to investigate whether the model can provide new insight into how decision making is affected across psychiatric disorders. At the end phase of the project, several subprojects have been concluded an published in scientific journal, while others are in the final stages. The collaboration with Dr. Frank at Brown University has resulted in the publication of a methodological paper, which describes an analysis toolbox that can be used to increase insight into the underlying cognitive mechanisms of learning and decision making. In addition, the project leader has established collaboration with research groups at MIT and Harvard, in a project concerning how patients suffering from depression are affected in the conflict between choosing between alternatives that contain both positive and negative outcomes. In this work, which has resulted in one scientific article and another currently in revision at a scientific journal, we have found that people with depressive symptoms, to a lesser degree than their peers, use information about the positive outcomes to inform their choice. A potential consequence of such a reduced sensitivity could be that they achieve less positive outcomes. The project leader has in addition been involved in the analysis of similar experimental tasks with research groups at Stanford University and Oslo University Hospital. Additional subprojects will be concluded after the end of the project period. These include the prediction of medication response in ADHD, and the development of a new model to describe the underlying processes of attention. These works have the potential to increase our understanding of how attention is affected in mental disorder.

Prosjektets mål var å utvikle nye metoder for å beskrive beslutningsprosesser, gjøre disse metodene tilgjengelig for andre forskere, samt anvende metodene for å bedre forstå hvordan mennesker med mentale lidelser er påvirket i prosessene metoden beskriver. Et annet mål har vært å etablere samarbeidprosjekter med forskningsgrupper i USA og Norge. De ovennevnte målene har i stor grad blitt oppnådd, og det er i tillegg flere mål som vil bli oppnådd etter prosjektets sluttfase, da flere arbeider er i ferd med å sluttføres. Vi har utviklet et metodeverktøy som gjør det lettere for andre forskere å anvende våre metoder. Ved å gjøre dette tilgjengelig for andre vil en effekt kunne være en økt forståelse av beslutningsprosesser, som kan øke forståelsen av avvik i beslutningsprosesser i mentale lidelser, og hvordan disse kan behandles. Prosjektet har lykkes med å etablere internasjonalt (og nasjonalt) forskningssamarbeid, inkludert forskningsgrupper på Brown, Harvard og MIT.

ADHD is associated with negative functional outcomes. The primary treatment of ADHD is stimulant medication, which alleviates symptoms and ameliorates negative life outcomes associated with the disorder. Despite the overall benefits of stimulant medication, about 2 of 3 adult patients with ADHD do not experience positive effects and therefore discontinue medication treatment. Early detection of these non-responders could greatly improve treatment of this subgroup and reduce costs associated with the use of medication. As to date, no neurobiological or behavioral results have enabled detection of this group. We have taken a novel computational modeling approach to this problem. Preliminary results from drift diffusion modeling (DDM) of the Continuous Performance Test (CPT), a commonly used neuropsychological test of attention, from 250 adult ADHD patients tested before and after start of medication treatment, indicate that we are able to separate medication responders and non-responders. This approach thus has the potential to be a first description of the mechanisms that predict medication response in ADHD, which can reduce the time to detect individual response to medication during the crucial phase after diagnosis. Through close collaboration with Dr. Michael Frank at Brown University, a leading figure in the field of computational psychiatry and developer of the most popular DDM software package, I plan to both implement and make accessible the methodology for analyzing CPT-results with the DDM, and use this approach to predict medication response in ADHD. During the final year of this proposed project I will, together with researchers at UiO and OUS, use the developed model to compare decision making deficits between clinical groups and relate these to their neurobiological underpinnings. The project has the potential to develop close connections between the research communities at Brown and in Oslo and greatly benefit my future research career.