Ateneo de Manila University

Low-Rank Adaptation (LoRA) significantly reduces compute and memory costs for finetuning Deep Learning models but is often harder to tune than dense training: when using factor-wise optimizers such as AdamW, it is sensitive to initialization choices, its optimal learning rates transfer poorly across ranks, and it often fails to beat dense baselines. We derive LoRA-Muon by applying the Muon optimizer's spectral steepest-descent rule to the low-rank setting. Along with our split weight-decay rule, our main claim is that LoRA-Muon is a good low-rank proxy for full-rank Muon and Shampoo-family optimizers. Its optimal learning rates transfer across rank, width, depth, and factor-rescaling. In our compute-matched TinyShakespeare study, a rank-2 proxy recovers the dense best tested learning rate, and a rank-32 LoRA-Muon run attains lower mean validation loss than the dense baseline in the seed-averaged sweep. We further show that the Spectron optimizer depends on arbitrary factor scaling, so it would likely be a poor fit when finetuning starts from badly imbalanced factors, and that LoRA-RITE's simplified QR-coordinate core implements the same spectral update. LoRA-Muon computes that update without QR-decomposition and avoids storing second moments, making it more accelerator-friendly and memory-efficient.

Background: Mental health concerns have been steadily rising worldwide, including in the Philippines. Given the shortage of mental health specialists in the country, it is crucial to make psychosocial support more accessible and available to Filipino youth.Objective: This study evaluated the feasibility of implementing Katatagan Plus, a resilience intervention for youth at risk of mental health issues, in seven local government units across two regions in the Philippines. Specifically, it aimed to measure the effectiveness of the Katatagan Plus program in developing resilience through adaptive coping, socio-emotional skills, and well-being, and reducing the risk of substance use and depressive symptoms.Method: Seventy-one adolescents screened as being at risk for mental health problems were invited to participate in the program. Most (68%) were participants in community-based programs facilitated by municipal health officers, whereas the remaining (32%) participated in school-based programs facilitated by school guidance personnel. Pre- and post-test surveys were conducted to measure program efficacy and changes in mental health outcomes.Results: Significant improvements in adaptive coping and well-being were observed from the pre- and post-tests. The proportion of participants with scores indicating good well-being rose from 49% to 70% before and after the program. There was also a significant decrease in depression symptoms and substance use. No differences were found in terms of setting/service provider for adaptive coping, substance use, and depression. However, improvements in well-being were greater among students in school-based settings compared to community settings.Conclusion: Results suggest the potential of this early mental health intervention for young people at risk, as well as the need for referrals for those who may need more specialised support.

BACKGROUND:Global inequities in access to cancer diagnostics and treatment contribute to wide variation in cancer mortality-to-incidence ratios (MIRs), a proxy for survival. We aimed to develop an interpretable machine learning framework to quantify country-specific health system contributors to MIR and inform policy prioritization. MATERIALS AND METHODS:We assembled national MIRs from GLOBOCAN 2022 for 185 countries and health system indicators from multilateral sources, including gross domestic product (GDP) per capita, universal health coverage (UHC) index, radiotherapy centers per population, health spending (%GDP), out-of-pocket expenditure, work force densities (physicians; nurses/midwives; surgical work force), pathology availability, Human Development Index, and gender inequality index. A CatBoost gradient-boosting model was trained with repeated leave-one-country-out cross-validation (10 repeats; 1850 predictions). Nested hyperparameter optimization and strict leakage control were used. Model interpretability employed SHapley Additive exPlanations (SHAP; TreeExplainer) to generate global and country-level feature attributions. SHAP values, model-derived metrics quantifying each factor's contribution to cancer outcomes, were generated. Performance metrics included R2, root mean squared error (RMSE), mean absolute error, and Pearson correlation; uncertainty was estimated by bootstrap resampling. RESULTS:The model showed strong out-of-sample performance [R2 = 0.852, 95% confidence interval (CI) 0.801-0.891; RMSE 0.057, 95% CI 0.050-0.064]; correlation between predicted and observed MIRs was r = 0.923 (P = 8.30 × 10-78). Global SHAP contributions ranked GDP per capita (22.5%), radiotherapy centers per population (15.4%), and UHC index (12.9%) as the leading determinants. Country-specific SHAP profiles revealed substantial heterogeneity in dominant drivers across settings, enabling tailored policy levers (e.g. infrastructure, coverage expansion, or financial protection). An accompanying web interface provides country-level SHAP summaries for decision support. CONCLUSIONS:An explainable machine learning approach accurately predicts national MIRs and decomposes predictions into country-specific health system attributions. While ecological and noncausal by design, the SHAP profiles translate population-level associations into actionable hypotheses for prioritizing investments-highlighting, across many contexts, radiotherapy capacity and UHC expansion as recurrent levers, and underscoring that higher total health spending alone may be insufficient without strategic allocation. Prospective, country-specific evaluations are warranted to test whether targeting model-identified drivers improve cancer outcomes.

Cancer is a leading cause of death and its incidence is rising among younger populations, yet formal cancer education is rarely included in U.S. high school curricula. Adolescence is a critical period for establishing lifelong health behaviors, and emerging evidence suggests that school-based cancer education can improve knowledge, shift attitudes, and promote risk-reducing behaviors. In this narrative review, we summarize studies of high school and middle school cancer education programs, including curricula focused on cancer biology, modifiable risk factors (e.g., tobacco, alcohol, diet, physical activity, ultraviolet exposure), HPV vaccination, and symptom awareness. Across diverse settings, brief classroom interventions, interactive modules, and service-learning projects consistently increased cancer literacy, recognition of warning signs, and intentions to adopt preventive behaviors, with some reports of early behavior change. We further highlight the potential of cancer education to address health inequities by improving health literacy, reducing stigma, and fostering trust in the medical system, particularly in communities with limited access to care or who are at significant risk of financial toxicity. Finally, we outline pragmatic strategies for integrating cancer content into existing science, health, and social science courses; leveraging electives and clubs; and partnering with healthcare professionals and community organizations. Integrating cancer education into high school represents a feasible, equity-promoting strategy to reduce future cancer risk.

The maturity mismatch paradox of investment and financing has led to the inefficient upgrading of carbon-intensive industries, and fintech in the banking system has the potential to solve this problem. By utilizing the loan information of Chinese listed companies and banks’ digital transformation index to construct a firm-level banking fintech index, this study aims to investigate the effect of banking fintech on the innovation of firms in carbon-intensive industries. Empirical results show that banking fintech can significantly boost corporate innovation and help carbon-intensive industries address the financing challenges of transformation and innovation. The results remain robust after the use of instrumental variables, dynamic panel models, and quasi-natural experimental methods to account for potential endogeneity. The enabling effect of banking fintech operates through dual pathways: in the quantitative dimension, it helps enhance corporate financial resilience and alleviate financing constraints; in the qualitative dimension, it facilitates breaking down information barriers between banks and enterprises while directing capital flows toward high-potential innovation projects. Additionally, the impact of banking fintech in promoting innovation is greater for large enterprises and state-owned enterprises. Implications and recommendations for financial policies and fintech policies for carbon-intensive industries are presented.