The precise detection of biomarkers is crucial for early disease diagnosis, prognosis evaluation, and personalized treatment [1], [2], [3], [4], [5], [6]. Biomarkers such as proteins, nucleic acids, carbohydrates, lipids, various metabolites, and hormones hold significant value in disease diagnosis [7], [8], [9], [10], [11], [12]. The precise detection of biomarkers is crucial for disease diagnosis and treatment. However, conventional nanozymes struggle to meet the demands of highly sensitive detection due to their insufficient catalytic activity, poor selectivity, and low environmental stability [3], [4], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22].

High-entropy nanozymes (HENs) break through the limitations of conventional nanozymes by forming high-entropy solid-solution structures through the atomic-level homogeneous mixing of five or more metal elements, thereby achieving multi-element synergistic effects [23], [24], [25], [26]. In the 1990s, researchers created alloys with high mixing entropy by adding alloying components. In 2004, Ye et al. first synthesized multi-principal component alloys, which they named high-entropy alloys (HEAs) (Fig. 1) [27], [28], [29], [30]. Since the introduction of HEAs in 2004, they have been a major focus of research. The concept of HEAs has been extended to the nanozyme field, where five or more elements are uniformly mixed in a single solid-solution phase in equiatomic or near-equiatomic ratios. HENs exhibit unique electronic structures and synergistic effects, endowing them with excellent catalytic activity, stability, and tunability [31], [32], [33], [34].

The unique electronic coupling and lattice strain effects of HENs not only significantly enhance the density of active sites but also enable precise modulation of catalytic activity, such as optimizing the d-band center to strengthen substrate adsorption energy or integrating multi-enzyme activities through a variable-valence metal network [35], [36]. The disordered crystal structure of HENs endows them with exceptional chemical stability, maintaining activity even under extreme conditions [37], [38]. Moreover, their surface-abundant coordination sites facilitate efficient coupling with biomolecules, while the incorporation of photothermal or magnetic elements allows the construction of "detection-therapy" integrated smart platforms [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49]. This provides novel insights for advancing point-of-care diagnostic technologies.

As shown in Table 1, ML-assisted HENs exhibit revolutionary advantages over conventional nanozymes and natural enzymes. By leveraging ML algorithms to optimize multi-element compositions and atomic configurations, HENs achieve exceptional catalytic activity (e.g., 3–5 times higher turnover rates than natural horseradish peroxidase) and broad-spectrum substrate adaptability (unlike single-function natural enzymes) [28], [50], [51], [52], [53], [54]. Their entropy-stabilized structures demonstrate remarkable stability under extreme conditions (pH/temperature), overcoming the fragility of natural enzymes and the tunability limitations of conventional nanozymes.

ML-driven high-throughput screening further enables precise regulation of enzyme-mimicking properties (oxidase/peroxidase/superoxide dismutase-like activities), while stimulus-responsive designs allow real-time sensing modulation - a capability unattainable with static natural enzymes or traditional artificial enzymes [24], [55], [56], [57], [58]. The synergistic integration of adaptive catalysis, programmable functionality, and robust stability makes ML-designed HENs a next-generation biosensing platform.

This article reviews the latest research advancements of HENs in biomarker detection, focusing on their structural simulations, design strategies, recognition mechanisms, and practical applications in disease diagnosis. Finally, we discuss the challenges and provide prospects to offer new insights for the development of more efficient and precise biomarker detection methods.