Publications

Research, with the point of each paper up front.

Every paper below carries a one- or two-line "why it matters". Grouped by theme; filter to narrow. Prefer to explore visually? See it all as an interactive map ↗

Multimorbidity & health networks

Disease and symptom networks, comorbidity, and higher-order structure in ageing cohorts.

PLOS Computational Biology · 2026

First author

Interpreting higher-order dependence in multimorbidity using cohort data: A partial information decomposition approach

Hourican, C., Peeters, G., Melis, R.J.F., Kok, A., van Schoor, N.M., Wezeman, S., Lees, M., Olde Rikkert, M.G.M., Quax, R.

In older adults, some symptoms, signs and behaviours only reveal their effect on health when they occur together rather than one at a time. This work introduces an open, bias-aware workflow — partial information decomposition plus a "BUST" map — that detects and interprets these "together-only" synergistic combinations across a large ageing cohort, flagging feature pairs that conventional association measures overlook.

Age and Ageing · 2026

Joint first author

Symptoms provide predictive information on daily functioning beyond signs and diseases in middle-aged and older adults, particularly those with multimorbidity

Peeters, G., Hourican, C., Lees, M., Quax, R., Melis, R., Kok, A., van Schoor, N.M., Olde Rikkert, M.

In older adults living with several chronic conditions, the symptoms people actually experience carry information about their day-to-day functioning that formal disease diagnoses miss — and adding symptoms sharply improves prediction of functional limitations. This supports a more symptom-oriented view of multimorbidity.

Psychological Medicine · 2025

A multilayer network analysis of cardiovascular–depression comorbidity reveals symptom-specific molecular biomarkers

Li, J., Bosch, J.A., Rydin, A.O., Hourican, C., Koloi, A., Tassi, S.C., Mishra, P.P., Mishra, B.H., Kähönen, M., Lehtimäki, T., Raitakari, O.T., Laaksonen, R., Keltikangas-Järvinen, L., Juonala, M., Quax, R.

Heart disease and depression often occur together, but the biology connecting them is poorly understood. Using a network approach on ~1,700 people (validated in the UK Biobank), the study pinpoints specific metabolites and lipids that biologically link particular depressive symptoms to cardiovascular measures — suggesting shared pathways that could inform prevention and treatment.

Frontiers in Systems Biology · 2023

First author

Understanding multimorbidity requires sign–disease networks and higher-order interactions, a perspective

Hourican, C., Peeters, G., Melis, R.J.F., Wezeman, S.L., Gill, T.M., Olde Rikkert, M.G.M., Quax, R.

Multimorbidity research usually counts diseases or looks at them in pairs, which hides how signs, symptoms and diseases actually combine. This perspective argues for networks that link signs, symptoms and diseases together and capture higher-order (synergistic) interactions as hypergraphs — showing, via a synthetic model, that pairwise thinking can miss the best intervention or produce unexpected "side-effects" a hypergraph reveals.

Information theory & higher-order interactions

Methods for detecting and estimating synergy, redundancy, and multi-way dependence.

Entropy · 2024

Bias in O-Information Estimation

Gehlen, J., Li, J., Hourican, C., Tassi, S.C., Mishra, P.P., Lehtimäki, T., Kähönen, M., Raitakari, O., Bosch, J.A., Quax, R.

O-information is a popular tool for detecting whether three or more variables interact synergistically or redundantly, but this paper shows that with too few data points the estimate is systematically skewed toward synergy — so many trios flagged as "highly synergistic" may be close to independent. It offers a partial bias correction and null-model benchmarking to guard against these finite-sample artefacts.

Entropy · 2024

First author

Efficient Search Algorithms for Identifying Synergistic Associations in High-Dimensional Datasets

Hourican, C., Li, J., Mishra, P.P., Lehtimäki, T., Mishra, B.H., Kähönen, M., Raitakari, O.T., Laaksonen, R., Keltikangas-Järvinen, L., Juonala, M., Quax, R.

Some meaningful patterns in data only emerge when several variables are considered jointly (synergy), but finding these combinations usually means checking an astronomically large number of possibilities. This paper introduces stochastic search strategies that locate synergistic sets efficiently without exhaustive enumeration, making the analysis practical on large biomedical datasets.

Cross-domain applications

Applying these tools to problems in other fields.

European Heart Journal – Digital Health · 2024

Predicting early-stage coronary artery disease using machine learning and routine clinical biomarkers improved by augmented virtual data

Koloi, A., Loukas, V.S., Hourican, C., Sakellarios, A.I., Quax, R., Mishra, P.P., Lehtimäki, T., Raitakari, O.T., Papaloukas, C., Bosch, J.A., März, W., Fotiadis, D.I.

Machine-learning models fed only routine, inexpensive inputs — standard laboratory markers plus age, sex and smoking status — can accurately flag early-stage coronary artery disease, and augmenting the training set with synthetically generated "virtual patient" data further improves accuracy. This points toward cheaper, less invasive early screening for heart disease.