Machine learning-couched treatment algorithms tailored to individualized profile of patients with primary anterior chamber angle closure predisposed to the glaucomatous optic neuropathy

Proteomic Analysis of Aqueous Humor Identified Clinically Relevant Molecular Targets for Neovascular Complications in Diabetic Retinopathy

Diabetic retinopathy (DR) is a leading cause of blindness in adults under 40 in the developed world, with a significant proportion progressing to vision-threatening stages such as proliferative diabetic retinopathy (PDR) and neovascular glaucoma (NVG). This study aims to explore the molecular mechanisms underlying the progression from non-proliferative DR (NPDR) to PDR and NVG, focusing on identifying potential biomarkers and therapeutic targets. Utilizing discovery-based proteomics, specifically label-free quantification (LFQ) and Tandem mass tag (TMT), we analyzed aqueous humor (AH) proteins obtained during cataract surgery or anterior chamber paracentesis from patients with NPDR, PDR, and NVG. Validation of marker candidates for each disease state was conducted using triple quadrupole (QQQ)-MS for targeted protein quantification. Our proteomic analysis identified 2,255 proteins and gene ontology analysis and functional annotation highlighted key biological processes implicated in DR, such as lens development, immune responses and lipid metabolism. Validation of potential biomarkers identified 20 proteins with significant concentration changes, including several candidates with diagnostic utility based on ROC curve analysis. Further investigation into clinical relevance revealed that crystallin gamma-S (CRYGS) is strongly associated with cataract severity, highlighting its role as a potential marker for ocular complications in DR. Importantly, we identified that the pathological factors driving DR progression have a much greater impact than age, a previously known variable, in shaping the proteomic landscape of aqueous humor. Additionally, proteins associated with macular degeneration (CA1, CA2, and HBA1) were uncovered, providing new insights into overlapping mechanisms between DR and other retinal diseases. Finally, proteins linked to panretinal photocoagulation (PRP) treatment, including APOB and CST6, were identified, suggesting their involvement in the therapeutic response and post-treatment adaptation. These findings underscore the potential of AH proteomics in uncovering predictive biomarkers and elucidating the molecular pathogenesis of DR and its complications.

Eye yoga for glaucoma: recovery of vascular dysregulation and visual field function-a randomized controlled trial

Purpose

Because stress can aggravate vascular dysregulation (VD) in primary open-angle glaucoma (POAG), stress reduction by eye yoga (EY) was studied if this predictive, preventive, and personalized medical (3PM) approach could help normalize intraocular pressure (IOP), retinal vessel dynamics, and visual fields (VF).

Patients and methods

POAG patients were randomized to an EY (n = 15) or control group (n = 12). EY was practiced daily for 1 h for 1 month at home using an iPod-audio guide while control patients read relaxing books daily.

Results

After intervention, EY patients, but not controls, showed a 6.4% IOP reduction (p = 0.027) and had significant VF improvements (p < 0.001). After EY, pattern deviation recovered in VF regions where small microvessels showed reduced vasoconstriction (artery: p = 0.012; vein: p = 0.042) and improved mean artery diameter recovered significantly (p = 0.015). When pooling data of both groups, recovered VF regions, but not non-recovered fields, showed significantly larger arterial diameter gains (2.4 [- 0.3-5.3] MU) with no adverse events.

Conclusions and 3PM recommendations

Because EY reduces vasoconstriction and improves VF function in POAG, we propose the "eye ball retraction theory," whereby ocular muscle tension is induced by mental stress which is a contributing mechanism, or even the key mechanism, of POAG. Reducing stress by relaxation is therefore a remedy for it improves blood flow as the fundamental mechanism of vision recovery and restoration. VD reduction is therefore a valuable therapeutic target for glaucoma care and eye yoga home exercises are a safe and effective complementary 3PM method of POAG care.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00389-x.

Deciphering key roles of B cells in prognostication and tailored therapeutic strategies for lung adenocarcinoma: a multi-omics and machine learning approach towards predictive, preventive, and personalized treatment strategies

Background

Lung adenocarcinoma (LUAD) remains a significant global health challenge, with an urgent need for innovative predictive, preventive, and personalized medicine (PPPM) strategies to improve patient outcomes. This study leveraged multi-omics and machine learning approaches to uncover the prognostic roles of B cells in LUAD, thereby reinforcing the PPPM approach.

Methods

We integrated multi-omics data, including bulk RNA, ATAC-seq, single-cell RNA, and spatial transcriptomics sequencing, to characterize the B cell landscape in LUAD within the PPPM framework. Subsequently, we developed an integrative machine learning program that generated the Scissor+ related B cell score (SRBS). This score was validated in the training and validation sets, and its prognostic value was assessed along with clinical features to develop predictive nomograms. This study further assessed the role of SRBS and SRBS genes in response to immunotherapy and identified personalized drug targets for distinct risk subgroups, with gene expression verified experimentally to ensure tailored medical interventions.

Results

Our analysis identified 79 Scissor+ B cell genes linked to LUAD prognosis, supporting the predictive aspect of PPPM. The SRBS model, which utilizes multiple machine learning algorithms, performed excellently in predicting prognosis and clinical transformation, embodying the preventive and personalized aspects of PPPM. Multifactorial analysis confirmed that SRBS was an independent prognostic factor. We observed varying biological functions and immune cell infiltration in the tumor immune microenvironment (TIME) between the high- and low-SRBS groups, underscoring personalized treatment approaches. Notably, patients with elevated SRBS may exhibit resistance to immunotherapy but show increased sensitivity to chemotherapy and targeted therapies. Additionally, we found that LDHA, as an SRBS gene with significant clinical implications, may regulate the sensitivity of LUAD cells to cisplatin.

Conclusion

This study presents a B cell-associated gene signature that serves as a prognostic marker to facilitate personalized treatment for patients with LUAD, adhering to the principles of PPPM.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00390-4.

Health Risk Assessment Using Machine Learning: Systematic Review

According to the World Health Organization, chronic illnesses account for over 70% of deaths globally, underscoring the need for effective health risk assessment (HRA). While machine learning (ML) has shown potential in enhancing HRA, no systematic review has explored its application in general health risk assessments. Existing reviews typically focus on specific conditions. This paper reviews published articles that utilize ML for HRA, and it aims to identify the model development methods. A systematic review following Tranfield et al.’s three-stage approach was conducted, and it adhered to the PRISMA protocol. The literature was sourced from five databases, including PubMed. Of the included articles, 42% (11/26) addressed general health risks. Secondary data sources were most common (14/26, 53.85%), while primary data were used in eleven studies, with nine (81.81%) using data from a specific population. Random forest was the most popular algorithm, which was used in nine studies (34.62%). Notably, twelve studies implemented multiple algorithms, while seven studies incorporated model interpretability techniques. Although these studies have shown promise in addressing digital health inequities, more research is needed to include diverse sample populations, particularly from underserved communities, to enhance the generalizability of existing models. Furthermore, model interpretability should be prioritized to ensure transparent, trustworthy, and broadly applicable healthcare solutions.

The AI revolution in glaucoma: Bridging challenges with opportunities

Recent advancements in artificial intelligence (AI) herald transformative potentials for reshaping glaucoma clinical management, improving screening efficacy, sharpening diagnosis precision, and refining the detection of disease progression. However, incorporating AI into healthcare usages faces significant hurdles in terms of developing algorithms and putting them into practice. When creating algorithms, issues arise due to the intensive effort required to label data, inconsistent diagnostic standards, and a lack of thorough testing, which often limits the algorithms' widespread applicability. Additionally, the "black box" nature of AI algorithms may cause doctors to be wary or skeptical. When it comes to using these tools, challenges include dealing with lower-quality images in real situations and the systems' limited ability to work well with diverse ethnic groups and different diagnostic equipment. Looking ahead, new developments aim to protect data privacy through federated learning paradigms, improving algorithm generalizability by diversifying input data modalities, and augmenting datasets with synthetic imagery. The integration of smartphones appears promising for using AI algorithms in both clinical and non-clinical settings. Furthermore, bringing in large language models (LLMs) to act as interactive tool in medicine may signify a significant change in how healthcare will be delivered in the future. By navigating through these challenges and leveraging on these as opportunities, the field of glaucoma AI will not only have improved algorithmic accuracy and optimized data integration but also a paradigmatic shift towards enhanced clinical acceptance and a transformative improvement in glaucoma care.

Body fluid multiomics in 3PM-guided ischemic stroke management: health risk assessment, targeted protection against health-to-disease transition, and cost-effective personalized approach are envisaged

Because of its rapid progression and frequently poor prognosis, stroke is the third major cause of death in Europe and the first one in China. Many independent studies demonstrated sufficient space for prevention interventions in the primary care of ischemic stroke defined as the most cost-effective protection of vulnerable subpopulations against health-to-disease transition. Although several studies identified molecular patterns specific for IS in body fluids, none of these approaches has yet been incorporated into IS treatment guidelines. The advantages and disadvantages of individual body fluids are thoroughly analyzed throughout the paper. For example, multiomics based on a minimally invasive approach utilizing blood and its components is recommended for real-time monitoring, due to the particularly high level of dynamics of the blood as a body system. On the other hand, tear fluid as a more stable system is recommended for a non-invasive and patient-friendly holistic approach appropriate for health risk assessment and innovative screening programs in cost-effective IS management. This article details aspects essential to promote the practical implementation of highlighted achievements in 3PM-guided IS management.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00376-2.

Genetic association of lipid traits and lipid-related drug targets with normal tension glaucoma: a Mendelian randomization study for predictive preventive and personalized medicine

Background

Glaucoma is the leading cause of irreversible blindness worldwide. Normal tension glaucoma (NTG) is a distinct subtype characterized by intraocular pressures (IOP) within the normal range (< 21 mm Hg). Due to its insidious onset and optic nerve damage, patients often present with advanced conditions upon diagnosis. NTG poses an additional challenge as it is difficult to identify with normal IOP, complicating its prediction, prevention, and treatment. Observational studies suggest a potential association between NTG and abnormal lipid metabolism, yet conclusive evidence establishing a direct causal relationship is lacking. This study aims to explore the causal link between serum lipids and NTG, while identifying lipid-related therapeutic targets. From the perspective of predictive, preventive, and personalized medicine (PPPM), clarifying the role of dyslipidemia in the development of NTG could provide a new strategy for primary prediction, targeted prevention, and personalized treatment of the disease.

Working hypothesis and methods

In our study, we hypothesized that individuals with dyslipidemia may be more susceptible to NTG due to a dysregulation of microvasculature in optic nerve head. To verify the working hypothesis, univariable Mendelian randomization (UVMR) and multivariable Mendelian randomization (MVMR) were utilized to estimate the causal effects of lipid traits on NTG. Drug target MR was used to explore possible target genes for NTG treatment. Genetic variants associated with lipid traits and variants of genes encoding seven lipid-related drug targets were extracted from the Global Lipids Genetics Consortium genome-wide association study (GWAS). GWAS data for NTG, primary open angle glaucoma (POAG), and suspected glaucoma (GLAUSUSP) were obtained from FinnGen Consortium. For apolipoproteins, we used summary statistics from a GWAS study by Kettunen et al. in 2016. For metabolic syndrome, summary statistics were extracted from UK Biobank participants. In the end, these findings could help identify individuals at risk of NTG by screening for lipid dyslipidemia, potentially leading to new targeted prevention and personalized treatment approaches.

Results

Genetically assessed high-density cholesterol (HDL) was negatively associated with NTG risk (inverse-variance weighted [IVW] model: OR per SD change of HDL level = 0.64; 95% CI, 0.49-0.85; P = 1.84 × 10-3), and the causal effect was independent of apolipoproteins and metabolic syndrome (IVW model: OR = 0.29; 95% CI, 0.14-0.60; P = 0.001 adjusted by ApoB and ApoA1; OR = 0.70; 95% CI, 0.52-0.95; P = 0.023 adjusted by BMI, HTN, and T2DM). Triglyceride (TG) was positively associated with NTG risk (IVW model: OR = 1.62; 95% CI, 1.15-2.29; P = 6.31 × 10-3), and the causal effect was independent of metabolic syndrome (IVW model: OR = 1.66; 95% CI, 1.18-2.34; P = 0.003 adjusted by BMI, HTN, and T2DM), but not apolipoproteins (IVW model: OR = 1.71; 95% CI, 0.99-2.95; P = 0.050 adjusted by ApoB and ApoA1). Genetic mimicry of apolipoprotein B (APOB) enhancement was associated with lower NTG risks (IVW model: OR = 0.09; 95% CI, 0.03-0.26; P = 9.32 × 10-6).

Conclusions

Our findings supported dyslipidemia as a predictive causal factor for NTG, independent of other factors such as metabolic comorbidities. Among seven lipid-related drug targets, APOB is a potential candidate drug target for preventing NTG. Personalized health profiles can be developed by integrating lipid metabolism with life styles, visual quality of life such as reading, driving, and walking. This comprehensive approach will aid in shifting from reactive medical services to PPPM in the management of NTG.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00373-5.

Digital biomarkers: 3PM approach revolutionizing chronic disease management - EPMA 2024 position

Non-communicable chronic diseases (NCDs) have become a major global health concern. They constitute the leading cause of disabilities, increased morbidity, mortality, and socio-economic disasters worldwide. Medical condition-specific digital biomarker (DB) panels have emerged as valuable tools to manage NCDs. DBs refer to the measurable and quantifiable physiological, behavioral, and environmental parameters collected for an individual through innovative digital health technologies, including wearables, smart devices, and medical sensors. By leveraging digital technologies, healthcare providers can gather real-time data and insights, enabling them to deliver more proactive and tailored interventions to individuals at risk and patients diagnosed with NCDs. Continuous monitoring of relevant health parameters through wearable devices or smartphone applications allows patients and clinicians to track the progression of NCDs in real time. With the introduction of digital biomarker monitoring (DBM), a new quality of primary and secondary healthcare is being offered with promising opportunities for health risk assessment and protection against health-to-disease transitions in vulnerable sub-populations. DBM enables healthcare providers to take the most cost-effective targeted preventive measures, to detect disease developments early, and to introduce personalized interventions. Consequently, they benefit the quality of life (QoL) of affected individuals, healthcare economy, and society at large. DBM is instrumental for the paradigm shift from reactive medical services to 3PM approach promoted by the European Association for Predictive, Preventive, and Personalized Medicine (EPMA) involving 3PM experts from 55 countries worldwide. This position manuscript consolidates multi-professional expertise in the area, demonstrating clinically relevant examples and providing the roadmap for implementing 3PM concepts facilitated through DBs.

Multi-omic profiling reveals potential biomarkers of hepatocellular carcinoma prognosis and therapy response among mitochondria-associated cell death genes in the context of 3P medicine

Background

Cancer cell growth, metastasis, and drug resistance are major challenges in treating liver hepatocellular carcinoma (LIHC). However, the lack of comprehensive and reliable models hamper the effectiveness of the predictive, preventive, and personalized medicine (PPPM/3PM) strategy in managing LIHC.

Methods

Leveraging seven distinct patterns of mitochondrial cell death (MCD), we conducted a multi-omic screening of MCD-related genes. A novel machine learning framework was developed, integrating 10 machine learning algorithms with 67 different combinations to establish a consensus mitochondrial cell death index (MCDI). This index underwent rigorous evaluation across training, validation, and in-house clinical cohorts. A comprehensive multi-omics analysis encompassing bulk, single-cell, and spatial transcriptomics was employed to achieve a deeper insight into the constructed signature. The response of risk subgroups to immunotherapy and targeted therapy was evaluated and validated. RT-qPCR, western blotting, and immunohistochemical staining were utilized for findings validation.

Results

Nine critical differentially expressed MCD-related genes were identified in LIHC. A consensus MCDI was constructed based on a 67-combination machine learning computational framework, demonstrating outstanding performance in predicting prognosis and clinical translation. MCDI correlated with immune infiltration, Tumor Immune Dysfunction and Exclusion (TIDE) score and sorafenib sensitivity. Findings were validated experimentally. Moreover, we identified PAK1IP1 as the most important gene for predicting LIHC prognosis and validated its potential as an indicator of prognosis and sorafenib response in our in-house clinical cohorts.

Conclusion

This study developed a novel predictive model for LIHC, namely MCDI. Incorporating MCDI into the PPPM framework will enhance clinical decision-making processes and optimize individualized treatment strategies for LIHC patients.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00362-8.

Personalized Drug Therapy: Innovative Concept Guided With Proteoformics

Personalized medicine can reduce adverse effects, enhance drug efficacy, and optimize treatment outcomes, which represents the essence of personalized medicine in the pharmacy field. Protein drugs are crucial in the field of personalized drug therapy and are currently the mainstay, which possess higher target specificity and biological activity than small-molecule chemical drugs, making them efficient in regulating disease-related biological processes, and have significant potential in the development of personalized drugs. Currently, protein drugs are designed and developed for specific protein targets based on patient-specific protein data. However, due to the rapid development of two-dimensional gel electrophoresis and mass spectrometry, it is now widely recognized that a canonical protein actually includes multiple proteoforms, and the differences between these proteoforms will result in varying responses to drugs. The variation in the effects of different proteoforms can be significant and the impact can even alter the intended benefit of a drug, potentially making it harmful instead of lifesaving. As a result, we propose that protein drugs should shift from being targeted through the lens of protein (proteomics) to being targeted through the lens of proteoform (proteoformics). This will enable the development of personalized protein drugs that are better equipped to meet patients' specific needs and disease characteristics. With further development in the field of proteoformics, individualized drug therapy, especially personalized protein drugs aimed at proteoforms as a drug target, will improve the understanding of disease mechanisms, discovery of new drug targets and signaling pathways, provide a theoretical basis for the development of new drugs, aid doctors in conducting health risk assessments and making more cost-effective targeted prevention strategies conducted by artificial intelligence/machine learning, promote technological innovation, and provide more convenient treatment tailored to individualized patient profile, which will benefit the affected individuals and society at large.

[Application of artificial intelligence in glaucoma. Part 2. Neural networks and machine learning in the monitoring and treatment of glaucoma]

The second part of the literature review on the application of artificial intelligence (AI) methods for screening, diagnosing, monitoring, and treating glaucoma provides information on how AI methods enhance the effectiveness of glaucoma monitoring and treatment, presents technologies that use machine learning, including neural networks, to predict disease progression and determine the need for anti-glaucoma surgery. The article also discusses the methods of personalized treatment based on projection machine learning methods and outlines the problems and prospects of using AI in solving tasks related to screening, diagnosing, and treating glaucoma.

[Application of artificial intelligence methods in the diagnosis and treatment of primary angle-closure disease]

This article reviews literature on the use of artificial intelligence (AI) methods for the diagnosis and treatment of primary angle-closure disease (PACD). The review describes how AI techniques enhance the efficiency of population screening for anterior chamber angle closure, presents technologies utilizing deep learning, including neural networks, for the analysis of large datasets obtained through anterior segment imaging methods, such as anterior segment optical coherence tomography (AS-OCT), digital gonioscopy, and ultrasound biomicroscopy, and discusses methods for treating PACD with the help of AI. Integration of deep learning and imaging techniques represents a crucial step in optimizing the diagnosis and treatment of PACD, reducing the burden on the healthcare system.