{"id":"73c67bf1b24d","type":"article","url":"https://hartvaat.nl/2026/02/26/ai-clustering-onthult-drie-functionele-risicoprofielen-bij-hartfalen/","title":"AI-clustering onthult drie functionele risicoprofielen bij hartfalen","title_en":"Artificial intelligence-based clustering to identify functional risk phenotypes in heart failure","category":"algemeen","category_label":"Algemeen","professions":["cardioloog"],"tags":["vrouwen"],"journal":"Open Heart","doi":"http://openheart.bmj.com/cgi/content/short/13/1/e003530?rss=1","source_url":"https://doi.org/http://openheart.bmj.com/cgi/content/short/13/1/e003530?rss=1","authors":["Qiu","X.","Ma","J.","Xu","L.","Jiang","M.","Pu","J."],"significance":5,"published":"2026-02-26","source_date":"2026-02-26","image":"","kennis":["https://hartvaat.nl/kennis/cardiometabool/obesitas-en-hart/","https://hartvaat.nl/kennis/cardiometabool/diabetes-en-cardiovasculair-risico/"],"congress":"","summary_en":"Patients with heart failure (HF) often have undetected declines in cardiorespiratory fitness (CRF), increasing the risk of poor outcomes, but tools for early CRF risk stratification are lacking. In this study, unsupervised AI clustering on 15 multimodal variables (metabolic, inflammatory and body-composition measures) in 505 HF patients sought phenotypes, with external validation in 201 patients. Three phenotypes emerged: 'balanced', 'inflammatory-sarcopenic' and 'metabolically dysregulated'. Both non-balanced phenotypes had clearly higher odds of impaired CRF (peak oxygen uptake ≤20 mL/kg/min). The best predictive models reached an AUC of about 0.75. Combining multimodal data with AI clustering thus yields an interpretable model for early functional risk stratification — a basis for more targeted rehabilitation in heart failure.","created":"2026-07-03T10:32:25Z","updated":"2026-07-03T13:31:22Z","licence":"Citeer vrij, met bronvermelding en een link naar hartvaat.nl (de url van het record). Samenvattingen zijn redactioneel werk van HartVaat; de oorspronkelijke publicaties blijven van hun uitgevers (doi). Geen medisch advies.","body_markdown":"Patiënten met hartfalen (HF) hebben vaak een onopgemerkte achteruitgang van hun cardiorespiratoire fitheid (CRF), wat het risico op slechte uitkomsten verhoogt, maar instrumenten voor vroege CRF-risicostratificatie ontbreken. In deze studie werd met ongesuperviseerde AI-clustering op 15 multimodale variabelen (metabole, inflammatoire en lichaamssamenstellingsmaten) bij 505 HF-patiënten gezocht naar fenotypes; externe validatie volgde in 201 patiënten. Er werden drie fenotypes gevonden: 'gebalanceerd', 'inflammatoir-sarcopeen' en 'metabool ontregeld'. Beide niet-gebalanceerde fenotypes hadden een duidelijk hogere kans op een verminderde CRF (piek-zuurstofopname ≤20 ml/kg/min). De beste voorspellende modellen bereikten een AUC van circa 0,75. Door multimodale gegevens te combineren met AI-clustering ontstaat zo een interpreteerbaar model voor vroege functionele risicostratificatie — een aanknopingspunt voor gerichtere revalidatie bij hartfalen.","abstract_original":"Background\n

Patients with heart failure (HF) frequently suffer from undetected declines in cardiorespiratory fitness (CRF), which significantly increases their risk of poor outcomes. However, current clinical practice lacks effective tools for early CRF risk stratification.

\n\nMethods\n

We conducted an artificial intelligence (AI)-driven unsupervised clustering analysis based on 15 multimodal clinical variables—including metabolic, inflammatory and body composition indicators—in 505 patients with HF. The associations between clustering-derived phenotypes and CRF impairment (maximal oxygen uptake (VO2 max) ≤20 mL/kg/min) were evaluated using multivariable logistic regression and five supervised machine learning models. SHapley Additive exPlanations analysis was applied for model interpretability. External validation was performed in an independent cohort of 201 patients.

\n\nResults\n

Three distinct phenotypes were identified: balanced, inflammatory-sarcopenic and metabolically dysregulated. Compared with the balanced phenotype, both non-balanced phenotypes showed significantly higher odds of impaired VO2 max. In the derivation cohort test set, random forest (area under the curve (AUC)=0.75; 95% CI 0.62 to 0.87) and XGBoost (AUC=0.74; 95% CI 0.62 to 0.87) demonstrated the best discriminative performance. In the external validation cohort, the highest discrimination was observed for Naive Bayes (AUC=0.75; 95% CI 0.67 to 0.83), followed by random forest (AUC=0.74; 95% CI 0.58 to 0.91).

\n\nConclusion\n

By integrating multimodal clinical data with AI-driven clustering and machine learning, this study identified novel CRF risk phenotypes in patients with HF and established a highly interpretable and generalisable risk stratification model. These findings offer a valuable framework for early functional assessment and pave the way for precision rehabilitation strategies in HF management.

\n"}