Research

My current research centers around the use and improvement of data in healthcare. It is premised on the fact that there is a need for epistemological pluralism, that there are multiple valid ways of knowing, and is informed by the need for justice and equity in computing. This perspective recognizes the diversity in knowledge discovery, cognitive justice and its connection to equity in computing.

Some themes I am currently working on:

Health Data Integration

Mediverse: A schema and standard agnostic multimodal data integration framework based on knowledge graphs, large language models and uncertainty bounded prediction.

Real World Evidence

Models for Real World Evidence from healthcare data. This theme focuses on improvement of disease outcomes and early detection in highly uncertain contexts, especially where symptoms and features are unspecific.

Project(s)	Paper(s)
Small Cell Lung Cancer: Explainable and interpretable machine learning to assess survivability, prognostic risk stratification and explain the difference between clinical trials and real world data.	1, 2, 3, 4
FEMaLe: Finding Endometriosis using Machine Learning. Using patient reported lifestyle factors to identify triggers, patterns and phenotypes of Endometriosis.	5
Colorectal Cancer: Early Detection of Colorectal Cancer in Primary Care, based on multimodal EHRs from Swedish Primary Care. This focuses on detection of CRC before clinical suspicion, from highly uncertain features.	Forthcoming

Health Systems Improvement

Computational and participatory approaches towards improving the effectiveness of healthcare systems, accounting for clinical and logistical factors. These projects aim at data and model driven ways to improve policy, cinical and patient pathways, waiting times, resource utilization and so on.

Project(s)	Paper(s)
Data Driven Discovery: Process and sequence mining of patient pathways to identify bottlenecks in patient flows, discrepancies with clinical guidelines, inefficient resource utilization and so on. Spanning data science, machine learning and process mining, these approaches have been applied to large hospitals, primary care and youth psychiatry.	6, 7, 8, 9, 10
Systems Modeling: Systems dynamics models of factors that compose different systems. Models were automatically extracted using literature, validated by experts and modeled using network analysis and perturbation models. This was used to describe the mental health and wellbeing of children and young people, and used to improve programs and policies that promote wellbeing in this cohort.	11, 12, 13