Systematic review of multimodal machine learning models for heart failure classification and prognosis predictionCan combining data sources better predict heart failure outcomes than single data types alone?

Frontiers in Medicine Published April 16, 2026 DOI ↗ Editorial oversight: Dr. Amelia Tan, PhD · Internal Medicine & Chronic Disease

AI-generated summary of the cited source, checked by automated accuracy review. How we work

Key Takeaway

Note inconsistent reporting and lack of prospective validation in this systematic review of multimodal machine learning for heart failure.

This systematic review evaluates the utility of multimodal machine learning models compared to unimodal approaches in the context of heart failure. The analysis encompasses 15 included studies, though the specific population and setting were not reported. The primary focus was on the performance of these models for secondary outcomes including HF classification and prognosis prediction. No primary outcome was explicitly defined in the source text.

The synthesized findings indicate that multimodal models demonstrated superior performance relative to unimodal approaches. Specifically, area under the receiver operating characteristic curve values frequently exceeding 80% and reaching as high as 98.2% were observed. It is important to note that absolute numbers, p-values, and confidence intervals were not reported for these results. Furthermore, no safety data, adverse events, or discontinuations were documented in the included studies.

The authors identify several critical limitations that constrain the current evidence base. These include inconsistent reporting of performance metrics and their 95% confidence intervals, limited external validation, and a near absence of prospective studies. Additionally, there is a deficiency in integrating genetic or 'omics' information with conventional data. Funding sources and potential conflicts of interest were not reported.

Given the lack of prospective validation and inconsistent metric reporting, the clinical relevance of these findings remains uncertain. Practitioners should interpret these results with caution, recognizing that the evidence does not yet support the routine implementation of such models without further prospective investigation.

When someone has heart failure, predicting their future health is critical. A recent look at 15 different studies compared two ways of making these predictions: using a single type of data versus mixing many different data sources together. The mixed approach, called multimodal, frequently outperformed the single-source method. In many cases, the accuracy scores reached very high levels, sometimes hitting 98.2 percent. This suggests that looking at a patient's full picture might give doctors a clearer view than looking at just one piece of the puzzle.

However, there are important gaps in what we know right now. The studies included in this review did not always share their specific accuracy numbers or the range of confidence around those numbers. Furthermore, most of the data used came from looking back at past records instead of testing these models on new patients as they are treated. Because of this, we cannot yet say these tools are ready for every hospital.

The review also noted that these models often missed important genetic information that could help explain heart failure better. Until more studies test these tools in real-time with diverse patients, doctors should treat these promising results with cautious optimism. The technology is advancing, but we need more proof that it works safely and consistently before changing how we care for patients.

What this means for you:

Combining multiple data sources often predicts heart failure outcomes better than single data types, but more real-world testing is needed.

Study Details

Study typeMeta analysis

EvidenceLevel 1

PublishedApr 2026

View Original Abstract ↓

IntroductionHeart failure (HF) is a global medical condition marked by substantial morbidity, mortality, and healthcare costs with complex pathophysiology and variation in definitions. Machine learning (ML) has emerged as a promising approach to improve HF classification and risk prediction by leveraging various data sources. This study aims to present the current state-of-the-art multimodal ML models for HF classification and prognosis prediction, focusing on their modalities, performance, and clinical utility.MethodsFollowing PRISMA guidelines and registered with PROSPERO (CRD420250654631), this review searched across four electronic databases (November 2014 – November 2024) and identified 284 unique records, of which 15 were included in the final synthesis. The quality of the studies was evaluated using QUADAS-2 and QUAPAS.ResultsOur results showed that the two most common multimodal combinations were tabular-image and tabular-text. The algorithms of the models included convolutional neural networks for image data, transformer-based approaches for text, with well-known fused techniques (early, middle, late fusion). Overall, multimodal models demonstrated superior performance compared to unimodal approaches, achieving area under the receiver operating characteristic curve values frequently exceeding 80% and reaching as high as 98.2%.ConclusionDespite promising results, challenges include inconsistent reporting of performance metrics and their 95% confidence intervals, limited external validation, a near absence of prospective studies, and a deficiency in integrating genetic or 'omics' information with conventional data. These challenges must be addressed to promote clinical adoption and future researchSystematic Review Registrationhttps://www.crd.york.ac.uk/PROSPERO/view/CRD420250654631, identifier CRD420250654631.

Systematic review of multimodal machine learning models for heart failure classification and prognosis predictionCan combining data sources better predict heart failure outcomes than single data types alone?

Study Details

Meta-analysis finds systemic lupus erythematosus elevates coronary artery disease risk

People with lupus face higher heart disease risk but genetics show no direct link

Clinical research that matters. Delivered to your inbox.

Systematic review of multimodal machine learning models for heart failure classification and prognosis predictionCan combining data sources better predict heart failure outcomes than single data types alone?

More on Heart Failure

Study Details

Meta-analysis finds systemic lupus erythematosus elevates coronary artery disease risk

People with lupus face higher heart disease risk but genetics show no direct link

Clinical research that matters. Delivered to your inbox.

Related in Cardiology

From Other Specialties