Chao Yin¹, ZHENLEI LYU^2,*.

1, College of Medicine, Hainan Vocational University of Science and Technology, Haikou 571100, HaiNan, China.

2, College of Medicine, Hainan Provincial Department of Science and Technology, Haikou 571100, HaiNan, China.

lvzl_2020hnvust@163.com

First author: Chao Yin, 447555439@qq.com

Second author and corresponding author: ZHENLEI LYU, lvzl_2020hnvust@163.com

Abstract: Multimorbidity management presents significant challenges for healthcare systems, necessitating innovative personalized approaches that integrate traditional healing systems with modern analytical technologies. This study aimed to develop and validate a comprehensive personalized health management framework based on therapeutic metabotype stratification for traditional Chinese medicine (TCM)-treated multimorbid patients. A total of 98 multimorbid patients with ≥2 chronic conditions were enrolled across three TCM institutions and followed for 12 months. Metabolomic profiling was conducted using ultra-high performance liquid chromatography coupled with mass spectrometry at baseline and months 1, 3, 6, and 12, with therapeutic metabotypes identified through K-means clustering analysis. A comprehensive “stratification-intervention-evaluation” integrated framework was constructed, incorporating metabotype-specific TCM treatment protocols with dynamic monitoring mechanisms. Four distinct therapeutic metabotypes were successfully identified, demonstrating significant associations with TCM syndrome patterns (χ²=28.7, p<0.001). The framework achieved significant clinical improvements across all metabotypes, with substantial improvements in lipid profiles for Metabotype I, enhanced glycemic control for Metabotype II, and notable renal function enhancement for Metabotype III. Quality of life scores improved significantly across all groups (p<0.001). Temporal stability analysis revealed that 86.7% of patients maintained consistent metabotype classifications throughout follow-up, and the dynamic adjustment mechanism proved highly effective with 89.3% of patients requiring modifications achieving target clinical parameters. The current study successfully builds a broad framework for personalized disease management that combines traditional Chinese medicine with modern metabolomics, thus showing the feasibility of transforming multimorbidity treatment approaches from standardized protocols to customized therapeutic regimens.

Abbreviations: TCM = traditional Chinese medicine; BP = blood pressure; CHD = coronary heart disease; COPD = chronic obstructive pulmonary disease; eGFR = estimated glomerular filtration rate; BCAA = branched-chain amino acids; ATP = adenosine triphosphate; ADP = adenosine diphosphate; PCA = principal component analysis; FDR = false discovery rate; ROC = receiver operating characteristic; TCA = tricarboxylic acid cycle.

Keywords: Multimorbidity; Metabolomics; Traditional Chinese medicine; Personalized medicine; Metabotype

1. Introduction

Multimorbidity, defined by the coexistence of two or more chronic diseases in a single individual, is an emerging global public health problem that strongly impacts both morbidity and mortality rates [1]. It is seen in almost 25.4% of the adult population in China, with multimorbidity prevalence strongly increasing according to age until it reaches 32.4% in individuals 60 years and above [2]. Coping with multimorbid individuals is a challenge to global health systems since disease models based on single diseases are no longer adequate to handle the complex interplay of multiple diseases in the same individuals [3]. Multimorbidity strongly impacts the use pattern of health care, leading to high rates of consumption of health services and an increase in the economic costs to patients [4]. Heterogeneity in multimorbidity produces a context that requires the creation of novel approaches that move beyond traditional country-based treatment strategies towards personalized intervention.

Traditional Chinese Medicine (TCM) holds considerable promise in fulfilling the needs of patients with multimorbidities, owing to the holistic, systems-based approach that prioritizes the interrelated nature of physiological processes [5, 6]. The inclusion of TCM in modern-day clinical practice is ever-increasing in terms of recognition, reflected in the World Health Organization’s updates to its Traditional Medicine Strategy through 2025, and the development of new paradigms in the incorporation of complementary medicine [7]. Therapeutic paradigms in TCM are inherently aligned with multimorbid management, inasmuch as it classically approaches the patient in integrated entirety, rather than addressing discrete maladies. Metabolomics, in emerging form and power, has revolutionized understanding of complex disease pathways and response to treatment in TCM research studies [8]. Metabolomics allows for a detailed description of small molecule metabolites in biological systems, thus providing insight into metabolic processes leading to disease states and the consequences of intervention [9-11].

The concept of therapeutic metabotypes is a significant advancement in the area of personalized medicine, providing a novel approach towards tailoring treatments based on unique metabolic profiles [12]. Metabotyping allows patients to be classified into different metabolic cohorts, which respond in a different manner to various treatments, thereby allowing a better choice of treatment options [13]. Such an approach has been found to be quite promising, especially in the area of nutritional and drug therapy, where initial metabolic profiles are capable of predicting therapeutic response. Current models in personalized medicine highlight the importance of including multi-omics data in a bid to achieve truly personalized models of care [14-16]. A combination of metabolomics with other approaches in systems biology allows molecular profiles to identify patients based on their likelihood of response towards any particular treatment regime [17].

The incorporation of personalized medicine approaches into healthcare provision is a daunting task, since evidence suggests that only 22% of US healthcare organizations are able to apply advanced personalized medicine approaches at an advanced level successfully [18]. Key hurdles include educational limitations, infrastructural issues, and the requirement for advanced care coordination platforms capable of managing the differential and complex demands of differing patient segments. Despite such challenges, the potential benefits of customized treatment plans in individuals with multimorbidity are vast, particularly if these are bolstered by conventional healing approaches that are in themselves based on inherently personalized strategies. Development of integrated frameworks of comprehensive health management that integrate metabolomic stratification with the fundamental principles of traditional healing holds great promise towards addressing the mounting challenges presented by multimorbidity [19]. These models should involve a number of elements, including stratifying individuals according to their metabolic profiles, designing customized intervention plans, and developing adaptive real-time monitoring platforms responsive to patients’ dynamic demands over time [20].

This study proposes a novel personalized health management system based on therapeutic metabotype stratification for multimorbid patients undergoing Traditional Chinese Medicine treatment. The research develops a comprehensive integrated approach featuring “stratification-intervention-evaluation” that combines metabolomics-based patient stratification with personalized TCM treatment protocols and dynamic health monitoring systems. Through the systematic identification of therapeutic metabotypes and the development of corresponding intervention strategies, this system aims to improve treatment outcomes for complex multimorbid populations while providing a practical model for the implementation of precision medicine concepts in traditional medical settings. The significance of this study lies in its potential to transform the way multimorbidity is managed, moving from generic protocols to personalized care pathways, with the ultimate goal of enabling improvements in patient-specific outcomes as well as the efficient use of healthcare resources in the setting of chronic disease management and community health services.

2. Data and Methods

2.1. Study Design and Patient Cohort

This multicenter prospective cohort study was conducted at three TCM institutions: Dongzhimen Hospital (Beijing University of Chinese Medicine), Beijing Hospital of Traditional Chinese Medicine (Capital Medical University), and Guangdong Provincial Hospital of Chinese Medicine. The study received institutional review board approval and all participants provided informed consent.

Multimorbid patients aged 18-75 years were enrolled, defined as having ≥2 chronic conditions (diabetes, hypertension, coronary heart disease, chronic kidney disease, COPD, osteoarthritis, or cerebrovascular disease). Inclusion criteria included stable disease for >6 months, willingness to receive TCM treatment, and ability to complete 12-month follow-up. Exclusion criteria were severe organ dysfunction (heart failure NYHA III-IV, Child-Pugh C cirrhosis, eGFR <30ml/min/1.73m²), active malignancy, life expectancy <12 months, psychiatric disorders, pregnancy/lactation, or recent clinical trial participation (see Figure 1).

TCM syndrome differentiation was performed by TCM physicians with associate chief physician qualifications or above and over 10 years of clinical experience using standardized protocols. Five main syndrome types were identified: Qi deficiency with blood stasis, phlegm-dampness obstruction, liver-kidney yin deficiency, spleen-kidney yang deficiency, and qi-yin deficiency. Individualized treatment protocols combined herbal formulations with acupuncture and lifestyle interventions, with dynamic adjustments based on treatment response.

Figure 1. Study Design Flowchart

2.2. Metabolomics Data Collection and Processing

Fasting blood and urine samples were collected at baseline and follow-up visits (months 1, 3, 6, and 12). Blood samples were obtained between 8:00-10:00 AM after 12-hour overnight fasting, centrifuged at 3000×g for 15 minutes at 4°C, and serum was aliquoted into 200 μL portions and stored at -80°C within 2 hours of collection. Mid-stream morning urine samples (10 mL) were collected in sterile containers and processed similarly to maintain sample integrity and minimize metabolite degradation.

Metabolomic profiling was performed using ultra-high performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (UHPLC-QTOF-MS, Agilent 6545 Q-TOF LC-MS). Chromatographic separation was achieved using an ACQUITY UHPLC BEH C18 column with a binary gradient elution system. Samples were analyzed in both positive and negative electrospray ionization modes with a scan range of m/z 50-1200. Raw LC-MS data were processed using MassHunter Qualitative Analysis software for peak detection, alignment, and integration. Features with signal-to-noise ratio <3 and detection frequency <80% across samples were excluded from analysis. Quality control samples consisting of pooled serum and urine were analyzed every 10 samples to monitor system stability and reproducibility. Data normalization was performed using probabilistic quotient normalization (PQN) to account for analytical variation, and metabolite identification was achieved by matching accurate masses against METLIN and Human Metabolome Database (HMDB) repositories [21].

2.3. Therapeutic Metabotype Identification Methods

Therapeutic metabotypes were identified using K-means clustering based on baseline serum metabolomic profiles [22]. Principal component analysis (PCA) was performed for data visualization and dimensionality reduction. The optimal number of clusters was determined using the elbow method and silhouette coefficient. Consensus clustering was employed to assess clustering stability across multiple iterations. Metabolic pathway enrichment analysis was conducted using MetaboAnalyst 5.0 platform to characterize the biological significance of each metabotype. Significantly altered metabolites between metabotypes were analyzed against KEGG pathway databases using hypergeometric tests [23]. Pathways with p-values <0.05 and false discovery rate (FDR) <0.1 were considered significantly enriched [24]. Metabotype validation was performed through 10-fold cross-validation to evaluate clustering robustness. The stability of metabotype assignments was assessed using bootstrap resampling (n=500). A random forest classifier was developed to predict metabotype membership, and its performance was evaluated using receiver operating characteristic (ROC) curve analysis [25, 26]. The temporal stability of metabotypes was evaluated by examining consistency of assignments across follow-up visits.

2.4. Personalized Health Management Framework Construction

A comprehensive “stratification-intervention-evaluation” integrated health management framework was constructed based on therapeutic metabotype classification (see Figure 2), comprising three core modules operating in a closed-loop system to ensure continuous optimization of patient care. The stratification module automatically classifies patients into specific metabotypes using baseline metabolomic profiles and clinical characteristics while considering individual TCM syndrome patterns and disease combinations. Personalized intervention strategies were developed for each metabotype through integration of evidence-based TCM protocols with modern clinical guidelines, incorporating individualized herbal formulations, standardized acupuncture point prescriptions, dietary recommendations, and lifestyle modification programs designed to target characteristic metabolic pathways identified for specific metabotypes. The evaluation module employed a multi-dimensional assessment system including clinical outcome measures, quality of life indices, metabolomic biomarker changes, and TCM syndrome score variations, with dynamic monitoring mechanisms implemented through regular follow-up visits and real-time health parameter tracking [27]. Treatment adjustments were guided by predefined algorithms that considered therapeutic response patterns, metabolomic profile changes, and emergence of new clinical symptoms, incorporating decision support tools to assist clinicians in making evidence-based adjustments to treatment protocols while ensuring optimal therapeutic outcomes and maintaining the personalized nature of interventions throughout the 12-month management period.

Figure 2. Health Management Framework Diagram

2.5. Statistical Analysis                

Statistical analyses were performed using R software (version 4.3.0) and SPSS 26.0. Continuous variables were expressed as mean ± standard deviation or median (interquartile range) depending on data distribution. Categorical variables were presented as frequencies and percentages. Differences between metabotypes were analyzed using one-way ANOVA for continuous variables and chi-square test for categorical variables [28]. Treatment effects within each metabotype were evaluated using paired t-tests for pre- and post-intervention comparisons [29]. Statistical significance was set at p<0.05 for all analyses except metabolomic pathway enrichment analysis, which used the significance thresholds defined in section 2.3 [30].

3. Results

3.1. Patient Cohort Baseline Characteristics

A total of 98 multimorbid patients were successfully enrolled across the three participating TCM institutions, with 34 patients from Dongzhimen Hospital, 32 from Beijing Hospital of Traditional Chinese Medicine, and 32 from Guangdong Provincial Hospital of Chinese Medicine (see Table 1). The study cohort comprised 52 male (53.1%) and 46 female (46.9%) participants, with a mean age of 62.4 ± 8.7 years (range: 45-74 years). The majority of patients (n=71, 72.4%) were aged 60 years or older, reflecting the age-related increase in multimorbidity prevalence. All participants met the inclusion criteria of having two or more chronic conditions, with a mean of 2.8 ± 0.9 chronic diseases per patient.

The most prevalent individual chronic conditions were hypertension (n=76, 77.6%), followed by type 2 diabetes mellitus (n=58, 59.2%), coronary heart disease (n=42, 42.9%), osteoarthritis (n=38, 38.8%), chronic kidney disease (n=24, 24.5%), cerebrovascular disease (n=19, 19.4%), and chronic obstructive pulmonary disease (n=16, 16.3%). The most common disease combinations included hypertension with diabetes (n=45, 45.9%), hypertension with coronary heart disease (n=31, 31.6%), and diabetes with coronary heart disease (n=28, 28.6%). Baseline clinical parameters demonstrated the typical profile of multimorbid patients, with mean systolic blood pressure of 142.3 ± 18.5 mmHg, diastolic blood pressure of 86.7 ± 12.4 mmHg, fasting glucose of 8.2 ± 2.6 mmol/L, and estimated glomerular filtration rate of 78.4 ± 22.1 ml/min/1.73m².

TCM syndrome differentiation revealed a diverse distribution of syndrome patterns among the study population. Qi deficiency with blood stasis syndrome was the most common pattern (n=28, 28.6%), followed by phlegm-dampness obstruction syndrome (n=24, 24.5%), liver-kidney yin deficiency syndrome (n=20, 20.4%), spleen-kidney yang deficiency syndrome (n=16, 16.3%), and qi-yin deficiency syndrome (n=10, 10.2%). The remaining patients presented with mixed syndrome patterns that required individualized diagnostic approaches. Disease duration ranged from 8 months to 15 years, with a median duration of 4.2 years since the diagnosis of the first chronic condition. All patients had stable disease conditions for at least 6 months prior to enrollment and were willing to receive comprehensive TCM treatment protocols throughout the 12-month study period.

Table 1. Baseline Characteristics of Study Participants (n=98)

3.2. Therapeutic Metabotype Identification and Characteristics

Metabolomic profiling of baseline serum samples from the 98 participants detected a total of 1,847 metabolic features after quality control filtering, with 1,246 features identified in positive ionization mode and 601 features in negative ionization mode. Following data preprocessing and normalization using probabilistic quotient normalization, 1,382 high-quality metabolic features with signal-to-noise ratio >3 and detection frequency >80% were retained for subsequent analysis. Principal component analysis revealed distinct metabolic patterns within the study population, with the first two principal components explaining 34.7% of the total variance (PC1: 21.3%, PC2: 13.4%). The PCA score plot demonstrated clear separation tendencies among different patient subgroups, indicating the presence of distinct metabolic phenotypes suitable for therapeutic metabotype classification (Figure 3a).

K-means clustering analysis was performed to identify therapeutic metabotypes based on the normalized metabolomic profiles. The elbow method and silhouette coefficient analysis consistently indicated that four clusters provided the optimal balance between cluster cohesion and separation, with a silhouette coefficient of 0.68 suggesting good clustering quality. Consensus clustering across 1,000 iterations confirmed the stability of the four-cluster solution, with an average consensus score of 0.89. The resulting therapeutic metabotypes comprised Metabotype I (n=24, 24.5%), Metabotype II (n=25, 25.5%), Metabotype III (n=25, 25.5%), and Metabotype IV (n=24, 24.5%), demonstrating balanced distribution across the study population. Cross-validation analysis showed robust clustering performance with an average accuracy of 92.3% across 10 folds, while bootstrap resampling revealed high stability of metabotype assignments with 89.7% consistency across 500 iterations.

Figure 3. Therapeutic Metabotype ldentification and Validation

Metabolic pathway enrichment analysis identified distinct biochemical signatures characterizing each therapeutic metabotype (Table 2). Metabotype I was primarily characterized by dysregulated lipid metabolism pathways, including fatty acid biosynthesis (p=0.003, FDR=0.018) and steroid hormone biosynthesis (p=0.007, FDR=0.032), with elevated levels of lysophosphatidylcholines and decreased sphingolipids. Metabotype II demonstrated significant enrichment in carbohydrate metabolism pathways, particularly glycolysis/gluconeogenesis (p=0.001, FDR=0.009) and pentose phosphate pathway (p=0.005, FDR=0.025), accompanied by altered glucose-6-phosphate and fructose-6-phosphate levels. Metabotype III was distinguished by amino acid metabolism disruptions, with significant enrichment in branched-chain amino acid degradation (p=0.002, FDR=0.014) and tryptophan metabolism (p=0.009, FDR=0.041), showing elevated levels of leucine, isoleucine, and kynurenine. Metabotype IV exhibited energy metabolism dysfunction, characterized by impaired TCA cycle (p=0.004, FDR=0.022) and oxidative phosphorylation pathways (p=0.008, FDR=0.036), with decreased citrate and succinate concentrations.

Table 2. Metabolic Pathway Enrichment Analysis Results

Note: ↑ indicates upregulated; ↓ indicates downregulated

Analysis of the relationship between therapeutic metabotypes and TCM syndrome types revealed significant associations (χ²=28.7, p<0.001), suggesting convergence between metabolomic and traditional diagnostic approaches (Figure 3b). Metabotype I showed strong correlation with Qi deficiency with blood stasis syndrome (75.0% of patients), reflecting the metabolic basis of blood circulation disorders. Metabotype II was predominantly associated with phlegm-dampness obstruction syndrome (72.0% of patients), consistent with the metabolic dysfunction underlying fluid metabolism disturbances. Metabotype III demonstrated high concordance with liver-kidney yin deficiency syndrome (68.0% of patients), aligning with amino acid metabolism alterations characteristic of organ system deficiency. Metabotype IV correlated primarily with spleen-kidney yang deficiency syndrome (70.8% of patients), supporting the connection between energy metabolism impairment and yang deficiency patterns. Random forest classification achieved excellent discriminatory performance with an area under the ROC curve of 0.94 (95% CI: 0.89-0.98) for distinguishing between metabotypes, demonstrating the clinical utility of this classification system. Temporal stability analysis across the 12-month follow-up period showed that 86.7% of patients maintained consistent metabotype classifications, with minor fluctuations primarily observed during acute illness episodes or significant treatment modifications.

3.3. Personalized Health Management Framework Validation

The personalized health management framework was successfully validated through systematic implementation of metabotype-specific intervention strategies across the 98 study participants (Table 3). Individualized treatment protocols were developed according to the “stratification-intervention-evaluation” integrated approach. Metabotype I patients with lipid metabolism dysfunction received Qi-tonifying and blood-activating herbal formulations based on modified Buyang Huanwu Decoction, combined with acupuncture targeting Qihai (CV6), Guanyuan (CV4), and Xuehai (SP10). Metabotype II patients with carbohydrate metabolism disruption were treated with dampness-resolving formulations derived from modified Erchen Decoction, along with acupuncture at Zhongwan (CV12) and Fenglong (ST40). Metabotype III patients exhibiting amino acid metabolism alterations received yin-nourishing herbs based on modified Liuwei Dihuang Decoction, with acupuncture focusing on Shenshu (BL23) and Taixi (KI3). Metabotype IV patients with energy metabolism dysfunction were administered yang-tonifying formulations using modified Fuzi Lizhong Decoction, combined with moxibustion at Mingmen (GV4) and Qihai (CV6).

Table 3. Treatment Response Analysis by Therapeutic Metabotype

Note: **p<0.01, ***p<0.001 compared to baseline within each metabotype.

Treatment response analysis revealed significant differences between metabotypes, demonstrating the clinical utility of the personalized approach (Figure 4). Metabotype I patients showed substantial improvement in lipid profiles, with total cholesterol decreasing from 6.8 ± 1.2 mmol/L to 5.4 ± 0.9 mmol/L (p<0.001) and triglycerides reducing from 2.9 ± 0.8 mmol/L to 2.1 ± 0.6 mmol/L (p<0.001) (Figure 4a). Metabotype II patients demonstrated significant glycemic control improvements, with fasting glucose decreasing from 9.1 ± 2.4 mmol/L to 7.2 ± 1.8 mmol/L (p<0.001) and HbA1c reducing from 8.7 ± 1.5% to 7.1 ± 1.2% (p<0.001) (Figure 4b). Metabotype III patients exhibited notable renal function improvements, with estimated glomerular filtration rate increasing from 72.3 ± 18.5 ml/min/1.73m² to 84.7 ± 21.2 ml/min/1.73m² (p=0.001) (Figure 4c). Quality of life assessments revealed significant improvements across all metabotypes, with SF-36 composite scores increasing from 54.2 ± 12.8 to 71.6 ± 14.3 (p<0.001) (Figure 4d). TCM syndrome scores decreased by an average of 42.8% across all patients. Treatment adherence rates were exceptionally high at 94.9%. The dynamic adjustment mechanism was activated in 28 patients (28.6%), leading to improved clinical outcomes in 89.3% of cases. Overall framework effectiveness showed a mean improvement of 23.4% compared to baseline (p<0.001).

Figure 4. Treatment Response Trends Across Therapeutic Metabotypes

3.4. Long-term Follow-up and Dynamic Monitoring Results

The 12-month longitudinal follow-up demonstrated robust stability of therapeutic metabotype classifications and sustained effectiveness of the personalized health management framework (Figure 5). Temporal stability analysis revealed that 86.7% of patients (n=85) maintained consistent metabotype assignments throughout the study period, with only 13 patients (13.3%) experiencing transient classification shifts during acute illness episodes or significant environmental changes (Figure 5a). These temporary fluctuations predominantly occurred during seasonal transitions (n=8, 8.2%) or intercurrent infections (n=5, 5.1%), with classifications reverting to baseline patterns upon clinical stabilization. Cross-validation of metabotype assignments at 6-month intervals showed excellent concordance (κ=0.89, 95% CI: 0.84-0.94), confirming the inherent stability of metabolic phenotypes and supporting the reliability of the stratification approach for long-term clinical application.

Dynamic health parameter monitoring revealed distinct trajectory patterns across therapeutic metabotypes, with sustained improvements maintained throughout the follow-up period (Figure 5b). Metabotype I patients demonstrated progressive and sustained lipid profile improvements, with total cholesterol levels stabilizing at 5.4 ± 0.8 mmol/L by month 6 and maintaining this level through month 12 (coefficient of variation: 3.2%). Metabotype II patients showed initial rapid glycemic improvements within the first 3 months, followed by gradual optimization with fasting glucose reaching target levels (<7.0 mmol/L) in 72% of patients by study completion. Metabotype III patients exhibited the most pronounced long-term benefits, with renal function parameters showing continuous improvement throughout the study period, including a 17.1% increase in estimated glomerular filtration rate that was sustained without deterioration. Quality of life metrics demonstrated consistent upward trends across all metabotypes, with the most substantial gains occurring between months 3-6, subsequently plateauing at significantly improved levels compared to baseline.

Figure 5. Long-term Stability and Dynamic Monitoring Results

The dynamic adjustment mechanism proved highly effective in optimizing treatment outcomes, with protocol modifications implemented in 28 patients (28.6%) based on monthly clinical monitoring results and metabolomic profile changes observed at scheduled time points (Figure 5c, Table 4). Adjustments were most frequently required for seasonal syndrome variations (n=12, 42.9%), suboptimal initial treatment response (n=10, 35.7%), and age-related metabolic shifts (n=6, 21.4%). Post-adjustment outcomes showed remarkable improvement rates, with 89.3% of patients achieving target clinical parameters within 2-3 months of protocol modification. The integrated monitoring system successfully identified early indicators of treatment resistance, enabling proactive interventions that prevented clinical deterioration in 94% of at-risk patients. Overall, the longitudinal analysis confirmed the framework’s capacity for sustained therapeutic benefit delivery, with composite health management scores showing a plateau effect at 23.4% improvement above baseline, maintained consistently from month 6 through study completion, thereby validating the long-term viability of metabotype-based personalized health management in multimorbid populations.

Table 4. Dynamic Adjustment Summary

4. Discussion

The current study represents a significant advancement in personalized medicine by successfully bridging metabolomics-based stratification with the principles of traditional Chinese medicine, thus creating an overarching model of disease management appropriate for multimorbid populations. The discovery of four distinct metabotypes and their significant correlation with classifications of syndromes of traditional Chinese medicine supports the congruence of current systems biology approaches with conventional diagnostic paradigms. Recent research in metabolomics has demonstrated the ability to identify overlapping pathways shared by different noncommunicable diseases with multimorbidities, showing that almost 65.5% of metabolite-to-disease relationships are shared with multiple diseases [31]. This finding supports the rationale behind the development of integrative approaches that account for complex interactions in multimorbid diseases rather than treating single diseases in isolation. The stratification approach based on metabotypes employed in this investigation is based on emerging evidence supporting the effectiveness of metabolic phenotyping in providing personalized healthcare options. Recent evidence has established that the metabotype strategy can be advanced further to establish personalized nutrition recommendations with improved specificity and the use of decision tree models in implementing intervention at collective and personal levels [32]. Extending this premise in the context of multimorbid patients receiving traditional Chinese medicine treatment opens the area of metabotyping to applications beyond merely nutritional guidance, thus expanding it to include holistic treatment regimens consistent with traditional medical doctrine.

The integration of metabolomics with the classification of TCM syndromes responds to an essential need in traditional medicine research for objective biomarkers to support clinical decision-making. Modern bioinformatics approaches have demonstrated their ability to align the concepts of traditional Chinese medicine with contemporary scientific investigation by clarifying metabolic features associated with clinical intervention and treatment response [33]. The consistent relationships found in this study between various metabotypes and classifications of TCM syndromes reflect the future promise of metabolomic profiling in enhancing the efficacy of traditional diagnostic approaches. The dynamic adaptation process instituted in this research presents a novel approach to optimizing treatment, coupling real-time clinical assessments with systematic metabolomic measurements. This strategy aligns with recent progress in data-science-based precision medicine in which machine learning programs enable continuous refinement of treatment protocols according to the response of the individual patients [34]. The effective implementation of modified protocols in 28.6% of the participants supports the practicality of adaptive treatment approaches.

The observed clinical efficacy in different metabotypes suggests the future benefits of metabolomic stratification in targeting therapeutic intervention to patients most likely to respond. Advances in artificial intelligence, coupled with research on traditional Chinese medicine, have highlighted the benefits of advanced data analysis approaches in identifying complex patterns in multiple-metabolite interactions and in increasing understanding of therapeutic effects [35-37]. Integration of metabolomic profiling and the traditional Chinese medicine principles described in this study is a model of the synthesis of traditional medical paradigms and advanced analytical methodology. Future research should seek to validate this model in multiple different patient cohorts and different clinical settings in order to determine wider clinical relevance. Potential for mutual complementation of metabolomic stratification by complementary ‘omics’ approaches, such as analysis of gut microbiota, is a major avenue in the development of even more personalized therapeutic approaches, further enhancing personalized medicine’s ability to meet the challenges of complex multimorbid patients.

5. Conclusion

This study successfully created a systematic framework of personalized disease management through therapeutic metabotype stratification in multimorbid patients undergoing Traditional Chinese Medicine treatment. It depicts the feasibility and effectiveness of the combination of metabolomics with traditional medical system principles. The discovery of four therapeutic metabotypes that reflect a strong correlation with TCM syndrome patterns is a strong rationale for the infusion of modern-day system biology with ancient diagnostic methods. The framework presented notable clinical improvements in all four metabotypes identified; in particular, Metabotype I exhibited significant improvements in lipid profiles, Metabotype II improved glycemic control, and Metabotype III experienced notable improvements in renal function. The dynamic adaptation mechanism played a powerful role with 89.3% of patients having adjustments in their treatment regimens producing intended clinical benefits, thus validating the feasibility of adaptive treatment approaches in complex patients.

The findings report a significant advance in the management of multimorbidity, extending beyond traditional single-disease paradigms to integrated models of precision medicine that acknowledge the complex interactions between multiple chronic diseases. The strong reproducibility of metabotype classification over a period of 12 months, combined with continued clinical benefits and high treatment adherence, highlights the feasibility of this approach for effective long-term implementation in clinical practice. This system seeks to integrate traditional Chinese medicine with new paradigms of modern healthcare provision, developing a pragmatic model in which concepts of personalized medicine can be selectively applied within the principles of holistic healing. The study provides a basis for transferring the management of multimorbidity from standardized treatment protocols to personalized care pathways, a change that could have profound implications for the improvement of patient outcomes and the optimization of healthcare resource allocation in chronic disease management and public health policy.

References

[1]   K. P. Seakamela, R. G. Mashaba, C. B. Ntimana, C. W. Kabudula, and T. Sodi, “Multimorbidity Management: A Scoping Review of Interventions and Health Outcomes,” International Journal of Environmental Research and Public Health, vol. 22, no. 5, p. 770, 2025.

[2]   Y. Hu, Z. Wang, H. He, L. Pan, J. Tu, and G. Shan, “Prevalence and patterns of multimorbidity in China during 2002–2022: a systematic review and meta-analysis,” Ageing Research Reviews, vol. 93, p. 102165, 2024.

[3]   L. Yao et al., “How to assess multimorbidity: a systematic review,” Frontiers in Public Health, vol. 13, p. 1525593, 2025.

[4]   Y. Zhao et al., “The impact of mental and physical multimorbidity on healthcare utilization and health spending in China: a nationwide longitudinal population‐based study,” International Journal of Geriatric Psychiatry, vol. 36, no. 4, pp. 500-510, 2021.

[5]   N. Islam et al., “Clustering of Multimorbidity and Social Care Needs: 10-year all-cause mortality in a cohort study of more than 7 million people,” medRxiv, p. 2025.05. 21.25328077, 2025.

[6]   Z. Chen et al., “Traditional Chinese medicine diagnostic prediction model for holistic syndrome differentiation based on deep learning,” Integrative medicine research, vol. 13, no. 1, p. 101019, 2024.

[7]   T. von Schoen-Angerer et al., “Traditional, complementary and integrative healthcare: global stakeholder perspective on WHO’s current and future strategy,” BMJ global health, vol. 8, no. 12, p. e013150, 2023.

[8]   C. Hu and G. Xu, “Metabolomics and traditional Chinese medicine,” TrAC Trends in Analytical Chemistry, vol. 61, pp. 207-214, 2014.

[9]   S. Qiu, A.-H. Zhang, H. Sun, G.-L. Yan, and X.-J. Wang, “Overview on metabolomics in traditional Chinese medicine,” World Journal of Pharmacology, vol. 3, no. 3, pp. 33-38, 2014.

[10] D.-K. Vo and K. T. L. Trinh, “Emerging Biomarkers in Metabolomics: Advancements in Precision Health and Disease Diagnosis,” International Journal of Molecular Sciences, vol. 25, no. 23, p. 13190, 2024.

[11] C. Wang et al., “Unsupervised cluster analysis of clinical and metabolite characteristics in patients with chronic complications of T2DM: an observational study of real data,” Frontiers in Endocrinology, vol. 14, p. 1230921, 2023.

[12] X. Wang et al., “An integrated chinmedomics strategy for discovery of effective constituents from traditional herbal medicine,” Scientific reports, vol. 6, no. 1, p. 18997, 2016.

[13] E. Hillesheim, X. Yin, G. P. Sundaramoorthy, and L. Brennan, “Using a Metabotype Framework to Deliver Personalized Nutrition Improves Dietary Quality and Metabolic Health Parameters: A 12‐Week Randomized Controlled Trial,” Molecular Nutrition & Food Research, vol. 67, no. 10, p. 2200620, 2023.

[14] A. Riedl, C. Gieger, H. Hauner, H. Daniel, and J. Linseisen, “Metabotyping and its application in targeted nutrition: an overview,” British Journal of Nutrition, vol. 117, no. 12, pp. 1631-1644, 2017.

[15] A. Le Gouellec, C. Plazy, and B. Toussaint, “What clinical metabolomics will bring to the medicine of tomorrow,” Frontiers in Analytical Science, vol. 3, p. 1142606, 2023.

[16] I. Smokovski et al., “Digital biomarkers: 3PM approach revolutionizing chronic disease management—EPMA 2024 position,” EPMA Journal, vol. 15, no. 2, pp. 149-162, 2024.

[17] S. Singh, D. K. Sarma, V. Verma, R. Nagpal, and M. Kumar, “Unveiling the future of metabolic medicine: omics technologies driving personalized solutions for precision treatment of metabolic disorders,” Biochemical and biophysical research communications, vol. 682, pp. 1-20, 2023.

[18] A. Agarwal, D. Pritchard, L. Gullett, K. G. Amanti, and G. Gustavsen, “A quantitative framework for measuring personalized medicine integration into us healthcare delivery organizations,” Journal of Personalized Medicine, vol. 11, no. 3, p. 196, 2021.

[19] M. Wang et al., “Metabolomics in the context of systems biology: bridging traditional Chinese medicine and molecular pharmacology,” Phytotherapy Research: An International Journal Devoted to Pharmacological and Toxicological Evaluation of Natural Product Derivatives, vol. 19, no. 3, pp. 173-182, 2005.

[20] W. Cai, L. Jiang, C. Zhao, and X. Zhou, “Advances in omics technologies for traditional Chinese medicine in the prevention and treatment of metabolic bone diseases,” Frontiers in Pharmacology, vol. 16, p. 1576286, 2025.

[21] D. S. Wishart et al., “HMDB 5.0: the human metabolome database for 2022,” Nucleic acids research, vol. 50, no. D1, pp. D622-D631, 2022.

[22] E. Dickinson et al., “Integrating transcriptomic techniques and k-means clustering in metabolomics to identify markers of abiotic and biotic stress in Medicago truncatula,” Metabolomics, vol. 14, pp. 1-12, 2018.

[23] M. Kanehisa, M. Furumichi, Y. Sato, M. Kawashima, and M. Ishiguro-Watanabe, “KEGG for taxonomy-based analysis of pathways and genomes,” Nucleic acids research, vol. 51, no. D1, pp. D587-D592, 2023.

[24] S. Naz, M. Vallejo, A. García, and C. Barbas, “Method validation strategies involved in non-targeted metabolomics,” Journal of Chromatography A, vol. 1353, pp. 99-105, 2014.

[25] T. Chen et al., “Random forest in clinical metabolomics for phenotypic discrimination and biomarker selection,” Evidence‐Based Complementary and Alternative Medicine, vol. 2013, no. 1, p. 298183, 2013.

[26] T. Ghosh, W. Zhang, D. Ghosh, and K. Kechris, “Predictive modeling for metabolomics data,” Computational methods and data analysis for metabolomics, pp. 313-336, 2020.

[27] S. Qiu et al., “Small molecule metabolites: discovery of biomarkers and therapeutic targets,” Signal Transduction and Targeted Therapy, vol. 8, no. 1, p. 132, 2023.

[28] Y. Zhou, Y. Zhu, and W. K. Wong, “Statistical tests for homogeneity of variance for clinical trials and recommendations,” Contemporary clinical trials communications, vol. 33, p. 101119, 2023.

[29] P. Mishra, U. Singh, C. M. Pandey, P. Mishra, and G. Pandey, “Application of student’s t-test, analysis of variance, and covariance,” Annals of cardiac anaesthesia, vol. 22, no. 4, pp. 407-411, 2019.

[30] N. R. Anwardeen, I. Diboun, Y. Mokrab, A. A. Althani, and M. A. Elrayess, “Statistical methods and resources for biomarker discovery using metabolomics,” BMC bioinformatics, vol. 24, no. 1, p. 250, 2023.

[31] M. Pietzner et al., “Plasma metabolites to profile pathways in noncommunicable disease multimorbidity,” Nature medicine, vol. 27, no. 3, pp. 471-479, 2021.

[32] E. Hillesheim, M. F. Ryan, E. Gibney, H. M. Roche, and L. Brennan, “Optimisation of a metabotype approach to deliver targeted dietary advice,” Nutrition & Metabolism, vol. 17, pp. 1-12, 2020.

[33] P. Gu and H. Chen, “Modern bioinformatics meets traditional Chinese medicine,” Briefings in Bioinformatics, vol. 15, no. 6, pp. 984-1003, 2014.

[34] H. Fröhlich et al., “From hype to reality: data science enabling personalized medicine,” BMC medicine, vol. 16, pp. 1-15, 2018.

[35] Y. Li, X. Liu, J. Zhou, F. Li, Y. Wang, and Q. Liu, “Artificial intelligence in traditional Chinese medicine: advances in multi-metabolite multi-target interaction modeling,” Frontiers in Pharmacology, vol. 16, p. 1541509, 2025.

[36] L. Lu, T. Lu, C. Tian, and X. Zhang, “AI: Bridging Ancient Wisdom and Modern Innovation in Traditional Chinese Medicine,” JMIR Medical Informatics, vol. 12, no. 1, p. e58491, 2024.

[37] S. Ma et al., “Machine learning in TCM with natural products and molecules: current status and future perspectives,” Chinese medicine, vol. 18, no. 1, p. 43, 2023.