Revolutionizing Artificial Intelligence-Driven Sepsis Detection in Critical Care: Clinical Translation of Exosome-Based Diagnostics for Sepsis – A Literature Review

INTRODUCTION

Sepsis is a dangerous organ failure brought on by a varying regulation of the host response to a normal infection. Despite breakthroughs in medical care for persons with a range of conditions, sepsis remains a major cause of severe cases and death among hospitalized patients.^[1]Disseminated intravascular coagulation (DIC), which exacerbates the illness, can also result from deregulation of coagulation pathways. To restore adequate blood flow and pressure, intravenous fluids are given. When an infection occurs, there are pathogen-associated molecular patterns as well as damage-associated molecular patterns, causing a dysregulated systemic inflammatory response that is mediated by pattern recognition receptors.^[2] Exosomes are naturally occurring extracellular vesicles (EVs) that range between 30 and 200 nm in diameter and they have garnered a lot of attention in recent times.^[3] Exosomes play a crucial role in the pathophysiology of sepsis because they transport miRNAs and other cargo that affect macrophage behavior and intercellular communication. These components found within exosomes are discharged into the surrounding environment after being integrated into the cell membrane. Exosomes are therefore essential for transmitting information and moving materials between different cells and organs. Exosomes have therapeutic potential for a variety of diseases. In an effort to find biomarkers linked to their complex pathophysiology, recent research has concentrated on comprehending the function of released exosomes in intricate pathogenic processes of sepsis.^[4]

With an emphasis on the function of microRNAs and their growing association with disease severity, this study consists of an updated overview of sepsis. In addition, we examine the pathophysiology of exosomes, their circulation during sepsis, and the state-of-the-art techniques for exosome isolation, transmission electron microscopy (TEM) visualization, and quantification. We also go over the diagnostic potential of exosomes, including CD36, tetraspanin profiles, and other promising biomarkers as well as why they might provide better sensitivity and specificity to help patients and clinicians by facilitating earlier detection, better risk assessment, and more focused treatment choices in critical care management.

MATERIAL AND METHODS

The literature searches were conducted using keywords such as exosome diagnosis, sepsis diagnosis, sepsis emergency, and intensive care sepsis. Our databases were gathered from websites such as Science Direct, PubMed, MDPI, and Google Scholar. The literature review was conducted from October 2025 to December 2025 by analyzing 81 articles that were retrieved from 2019 to 2025, but only 32 were reviewed after omitting articles that were duplicated and unsuitable for the topic. We collected and analyzed articles to form the informative literature to present as a narrative review.

RESULTS

Exosomes have a function in the immunosuppressive phase of sepsis; research has shown that exosomes formed from monocytes are downregulated and that exosomal microRNAs released from various cell types modulate immune-paralysis pathways. Increased levels of exosomal CD14 have been suggested as markers of higher mortality risk and further illustrate the systemic immunological imbalance typical of severe sepsis. This dual regulatory ability highlights the function of exosomes as important immune homeostasis modulators, serving as vital mediators in preserving or upsetting immunological balance during sepsis. Commercial isolation kits are a step in the right direction because they are easy to use, need little equipment, and work with a variety of biological samples, but they still have certain drawbacks. Their usefulness for extensive clinical application is hampered by their high cost, limited scalability, and variability in purity, yield, and size distribution.

The transition from earlier sepsis frameworks to Sepsis-3 has fundamentally shifted the clinical and biological phenotype of patient populations included in research studies. Sepsis-3 emphasizes life-threatening organ dysfunction caused by a dysregulated host response to infection and operationalizes this using the SOFA score, thereby focusing on clinically significant organ failure rather than broader inflammatory syndromes such as SIRS. This change means that earlier biomarker studies based on pre-Sepsis-3 criteria may have included heterogeneous or less severe patient cohorts, potentially diluting biomarker specificity and prognostic performance. In addition, the evolution from Sequential Organ Failure Assessment (SOFA)-1 to SOFA-2 reflects advances in organ support technologies, updated clinical practices, and improved global applicability, which can alter organ dysfunction classification and risk stratification. As a result, biomarker thresholds, predictive accuracy, and clinical utility may vary depending on the sepsis definition and scoring system applied. Future biomarker validation should therefore prioritize cohorts classified according to Sepsis-3 and SOFA-2 to ensure consistency, reproducibility, and clinical translatability, and stratified analyses should be performed when comparing historical datasets. Regarding exosomal miRNAs, the manuscript recognizes that many described miRNAs currently function as biomarkers reflecting underlying pathophysiological processes, but accumulating mechanistic evidence supports a subset as causal mediators in sepsis. For example, miR-155 has been shown to promote pro-inflammatory macrophage polarization and metabolic reprogramming through inhibition of SHIP1 and SOCS1 and activation of NF-κB signaling, indicating a direct role in amplifying inflammatory cascades and organ injury. Conversely, exosomal miR-146a exhibits anti-inflammatory effects by targeting interleukin-1 receptor-associated kinase 1 (IRAK1)/TNF receptor-associated factor 6 (TRAF6) and promoting macrophage M2 polarization, with therapeutic delivery improving septic cardiomyopathy and systemic inflammation in experimental models. Similarly, exosomal miR-145 modulates transforming growth factor-beta (TGF-β) signaling through transforming growth factor beta receptor 2 (TGFBR2) and has been shown to ameliorate sepsis-induced lung injury in vivo, supporting a mechanistic and potentially therapeutic role. These experimental gain- and loss-of-function studies, along with pathway analyses, provide evidence for causal involvement rather than mere association. In contrast, other miRNAs such as miR-150, miR-223, and miR-15a/16 have been primarily correlated with disease severity, immune dysregulation, and mortality and may represent epiphenomena reflecting immune activation, endothelial dysfunction, or metabolic stress rather than direct drivers of pathology. Distinguishing causal mediators from epiphenomena requires integrated approaches including mechanistic in vitro and in vivo models, longitudinal profiling, intervention studies, and systems biology analyses to confirm downstream biological effects. Overall, the manuscript emphasizes that exosomal miRNAs represent both biomarkers and biologically active mediators, and future translational studies integrating standardized sepsis definitions and mechanistic validation are essential to establish their clinical and therapeutic relevance.

Techniques for visualization can provide difficulties. Despite being widely utilized, fluorescent labeling techniques may not always be able to accurately detect exosomes because of problems with signal amplification associated with vesicle size and shape. By permitting direct visualization of particular exosomal proteins through antibody binding, immunogold TEM offers a more conclusive option that improves structural and molecular characterization accuracy. Exosomes have great potential as instruments for early diagnosis, risk assessment, and continuous monitoring in sepsis management due to their stability, accessibility, and biologically active cargo, as well as their dynamic changes during the septic process.

Exosome pathophysiology and circulation in sepsis

Exosomes are functional subtypes of EVs that form a complex network of communication in both systemic and localized molecular exchange. Typically, measuring a diameter of around 30–200 nm, they are generated from the endosomal pathway, starting out with the double invagination of the plasma membrane and consequent formation of multivesicular bodies containing intraluminal vesicles. This is then released into the extracellular environment as exosomes, which are ultimately enriched with a variety of biological fluids and a wide range of cargo, including surface proteins, intracellular enzymes, transcription factors, cytokines, lipids, metabolites, nucleic acids, and adhesion molecules. Thereby, endosomes play a major role in signaling pathways by affecting the physiological and pathological responses of recipient cells.^[12,15]

Most of all, exosome carried out transport has been of particular importance in the initiation, development, and treatment of inflammation in both local and systemic conditions. The diverse and contrasting effects exerted by exosomes have become especially relevant in sepsis. Vesicles are released from antigen- presenting cells such as dendritic cells, macrophages, and B lymphocytes which function mainly as immune activators. Therefore, they promote the maturation, differentiation, and expansion of various immune cell subsets, which in turn enhance both innate and adaptive responses against pathogens. Hence, exosomes function through dynamic regulations capable of accelerating or restraining immune activity, contributing a huge role in the host response to sepsis.^[12]

Hundreds of dysregulated proteins, mRNAs, and IncRNAs found in exosomes from septic patients indicate widespread alteration, increased release, and poor clearance during systemic inflammations like sepsis. Through the induction of the release of cytokines, NFkB signaling in macrophages activates and excites the release of high-mobility group box-1, and the hyperinflammatory state is further worsened. It has been demonstrated that certain exosomal microRNAs like miR-92a-3p can mediate crosstalk between alveolar cells, ALI as well as alveolar macrophage activation.^[15]

As understood, elevated exosomal CD14 levels, which represent the systemic immune dysregulation linked to advanced sepsis, have also been found to be potential markers of mortality risk.^[16,17] This dual and self-limited regulation that exosomes play acts as a key factor in the homeostasis of sepsis, maintaining a role as the regulator or balance switch for the immune system.^[12] [Figure 1].

Close

Figure 1:

Pathophysiology of inflammation: Regulation in sepsis. microRNA: Micro ribonucleic acid. NFκB: Nuclear factor kappa B, HMGB1: High mobility group box 1, MIR-92a-3p: microRNA-92a-3p, CD14: Cluster of differentiation 14.

Export to PPT

EXOSOME DIAGNOSTIC METHOD IN SEPSIS

DEPENDENT VARIABILITY OF miRNA EXPRESSION IN SEPSIS

Circulating miRNAs have emerged as valuable diagnostic and prognostic indicators in sepsis, particularly along the course of the manifestation clinically, reflecting its dynamic function in the nature of the disease and outcomes.^[29] In particular, miR-155-5p, miR-146a, miR-223, and miR-145 are found to demonstrate both reproducibility and context dependence across patient populations, with variability arising according to disease stage, immune marker association, and methodological differences. Across studies, miR-155-5p was consistently reported as a significant biomarker found in septic patients, correlating with disease severity. Elevated levels have been positively associated with higher SOFA scores, and ICU non-survivors had higher levels than survivors. In addition, circulating miR-155 decreases approximately 48 h after admission, limiting its severity in the early stage of sepsis. This finding could also however stem from inconsistencies across studies, whether it be when likely measured, highlighting inconsistent sampling windows leading to the variability in findings. Regardless, the expression of miR-155 is a great stepping stone as a marker of early inflammation and severity, as well as paving a way for its antagonistic derivation as a potential therapeutic approach.^[30]

Similarly, miR-146a has been shown to have a high diagnostic specificity and sensitivity, as evidenced by its elevated serum levels in septic patients compared to healthy controls. Thereby, miR-146a functions to regulate innate immune signaling by targeting IRAK-1 and TRAF-6, key mediators in TLRs and NF-κB pathways. The action as a negative feedback regulator of inflammation, leads to its increase in expression in early immune activation and fluctuation as immune suppression develops. Thus, the variability among infection sources, be it pneumonia, abdominal sepsis, or urosepsis, factors such as patient age or immune conditions can influence circulating levels, further highlighting its dynamic directionality.^[10] Higher serum levels of miR-223 are also seen in patient populations, and it has been linked to worse outcomes, such as organ failure and death. However, its expression varies in dependence on leukocyte activation states and inflammatory burden, reinforcing the importance of cohort composition. Additional miRNAs, such as miR-150 and miR-4772-5p, have been shown to demonstrate utility in differentiating SIRS from sepsis with high accuracy. This reiterates the strong distinguishing factors across miRNAs in infectious versus non-infectious inflammation, while others, miR-155 AND miR-146a, exhibiting phase dependent up and downs tied to immune activation and resolution.^[29] Overall, miR-155-5p, miR-146a, and miR-223 are continuously raised in septic populations, suggesting potential diagnostic performance. However, their magnitude and prognostic significance differ depending on the period of sample collection, immunological phase, infection source, and patient heterogeneity. Standardization of sample techniques and validation in larger, multicenter cohorts are critical for improving repeatability and enabling meaningful therapeutic translation of context-dependent miRNA biomarkers in sepsis [Figure 2].

Close

Figure 2:

Summary on exosome biogenesis, miRNA-mediated pathways, and clinical workflow integration. miRNA: Micro ribonucleic acid. TEM: Transmission electron microscopy, TLR: Toll-like receptor, SOFA: Sequential organ failure assessment, NfKb: Nuclear factor kappa-light-chain-enhancer of activated B cells, EM: Electron microscopy, NTA: Nanoparticle tracking analysis.

Export to PPT

NOVEL CONTRIBUTION COMPARED WITH EXISTING STUDIES

EVs play a vital role in intercellular communication, which recreates the physiological or pathological state of the cells they originate from. Although EVs’ clinical potential is becoming more widely acknowledged, the majority of the focus is now being driven by preclinical and biological research.^[31] In addition, the emerging topic of exosome research in sepsis offers exciting therapeutic potential. Blocking the release of harmful exosomes has improved outcomes in animal models, while providing beneficial exosomes, particularly from stem cells, has demonstrated multi-organ protective advantages.^[14] Nonetheless, a major problem for the industry is still the standardization and improvement of EV extraction, purification, and characterization procedures, which are still essential for repeatability and functional research later on.^[31]

A versatile team of data scientists, physicians, biologists, and laboratory specialists is required to address and overcome these obstacles. In the near future, the main focus should be on developing rapid and reproducible laboratory-level isolation and characterization methods. Moreover, since microenvironmental variations may change the activity of EV, it is burdensome to interpret study results from ex vivo, in vitro, and animal model studies into conditions that provide therapeutic advantage. Thus, it is necessary to focus on the development of quick and repeatable laboratory-level isolation and characterization techniques for future studies.^[32] Another main problem is about the exosome separation methods. Regularized techniques for exosome separation should be formulated to make sure the results are accurate to be used in therapeutic settings.^[4]

Surface-enhanced Raman scattering (SERS’s) efficacy is severely limited by the structural similarity of biological samples, which results in overlapping spectral profiles and makes it challenging to differentiate between various analytes and appropriately interpret the spectra. AI methods, especially machine learning (ML) and deep learning (DL), have become important tools to get over these restrictions. Large spectral datasets, complex patterns, and improved signal discrimination are all possible using artificial intelligence (AI) algorithms. AI-driven models can be used to automate SERS data analysis, increasing detection sensitivity, accuracy, and reproducibility. While DL techniques, such as convolutional neural networks, allow for more sophisticated feature extraction and interpretation, ML techniques, such as principal component analysis and support vector machines, aid in the classification of spectral data.^[33] A strong diagnostic model that demonstrated encouraging clinical prediction performance in sepsis was established thanks to the use of ML techniques.^[34] Exosome analysis using SERS and AI has great promise for use in biomedical applications, such as liquid biopsy development, tailored therapy, and early illness identification. SERS-based exosome detection can become a quick, dependable, and scalable technique for clinical and research contexts by utilizing AI, opening the door for novel therapeutic and diagnostic approaches.^[35,36]

This work offers information on the general function of exosomes in sepsis as well as fresh viewpoints on exosomal miRNA and its use in critical care treatment, prognosis, and diagnostics. In addition, our research indicates that liquid biopsies may be the most accurate way to assess the involvement of exosomes in diagnosing, treating, and anticipating sepsis due to their lipid bilayer. Exosomes have a lipid bilayer protection that also makes them strong in the blood, resistant to degradation, quicker than microbes, and easier to recognize.

CONCLUSION

With all the advances in critical care, sepsis moves ahead to drive high mortality rates in ICUs, largely due to the complex nature of the immune response and the limitations of current diagnostic timing. The pointers of this review show the transformative potential of exosomes as superior candidates for liquid biopsy. Unlike conventional markers such as procalcitonin, C-reactive protein, or presepsin, exosomes offer a unique advantage: Their lipid bilayer protects sensitive cargo including miRNAs and proteins, providing a stable and highly specific snapshot of the patient’s physiological state from hyperinflammation to immune paralysis and organ injury. The clinical utility of exosomal profiling is evident in its predictive power.