Southeast Asian Journal of Sciences

OPTIMIZATION OF TRUST PROPAGATION ON HILBERT SPACE IN COMPLEX NETWORKS

Dinh Que Tran — 2026-09-09

This paper introduces a Hilbert space formulation of trust propa-

gation and develops an optimization framework for estimating global

trust states in complex networks. Trust interactions are modeled using

a bounded linear operator defined on the Hilbert space of node trust

vectors. The global trust inference problem is formulated as a varia-

tional optimization problem whose solution represents the equilibrium

trust configuration of the network. The proposed framework guarantees

existence and uniqueness of optimal trust states under mild spectral con-

ditions. An eﬃcient iterative algorithm is derived using gradient-based

optimization in Hilbert space. Preliminary experiments on simulated

networks demonstrate the stability and convergence of the proposed ap-

proach compared with classical propagation models.

INITIAL STUDY ON HUMAN PSYCHOLOGY AND THE INFLUENCE OF COMMUNICATION IN SOCIAL NETWORKS

Ngoc Ha T. Tran — 2026-09-09

Human psychological states evolve through interactions with commu-

nication environments and social networks. This paper presents a con-

ceptual and mathematical framework describing how psychological states

change through communication interactions among individuals. Individ-

uals are modeled as nodes in a network whose psychological states evolve

through interpersonal influence and external information exposure. Two

theoretical propositions analyze convergence properties of the model, and

a simulation experiment illustrates psychological dynamics in a small net-

work.

SPD-SKINNET: SPECTRAL RIEMANNIAN LEARNING WITH COVARIANCE REPRESENTATIONS FOR SKIN LESION CLASSIFICATION

Dinh Que Tran — 2026-09-09

Automatic skin lesion diagnosis is a fundamental challenge in medi-

cal image analysis because of requiring accurate and interpretable rep-

resentations of complex visual patterns. Most existing deep learning

approaches rely on Euclidean feature embeddings, which primarily cap-

ture first-order statistics and often fail to model the intrinsic geometric

and textural structure of skin lesions. In this paper, we propose SPD-

SkinNet, a spectral–Riemannian deep learning framework that repre-

sents skin lesion features as symmetric positive definite (SPD) matrices,

enabling the modeling of second-order statistics via feature covariance.

By performing spectral decomposition on the SPD manifold, we extract

eigenvalues that serve as compact and interpretable descriptors of lesion

texture. The initial experimental results on the synthetic highlight the

potential of integrating Riemannian geometry and spectral analysis into

deep learning for medical imaging.

THE STUDY FOR MEDICAL DEVICE RELATED PRESSURE INJURY IN CRITICAL CARE UNIT IN VIETNAM

Tran Thi Thuy Hang — 2026-09-09

Medical device-related pressure injuries (MDRPIs) are a significant

concern in Intensive Care Units (ICUs) due to the extensive use of medi-

cal devices. These injuries can lead to prolonged hospital stays, increased

healthcare costs, and diminished quality of life for patients. Despite the

prevalence of MDRPIs, there is limited research on this issue in Vietnam.

This study aims to investigate the incidence, level of injury and risk

factors associated with MDRPIs in critically ill patients at the University

Medical Center in Ho Chi Minh City, Vietnam.

A descriptive study was conducted from January to April 2024, in-

cluding 146 ICU patients who met the inclusion criteria. Data on demo-

graphics, medical device use, and the incidence of MDRPIs were collected

based on days period in ICU. Statistical analyses, including univariate

and multivariate logistic regression, were performed to identify significant

risk factors.

PARENTIAL KNOWLEDGE, PERSPECTIVE AND CAREGIVING SKILLS FOR CHIDREN WITH CONGENTIAL HEART DISEASE IN VIETNAM

Dang Van Sang — 2026-09-09

Background: This study examined the relationship between disease

awareness, disposition, and caregiving capacity in parents of children

with congenital heart disease in selected Vietnamese communities.

Methods: A multi-phase study was conducted with 275 parents us-

ing simple random sampling. Statistical analyses included frequency,

percentage, mean calculation, Pearson’s correlation coeﬃcient and re-

gression analysis.

Results: Parents demonstrated high awareness of prevention but

moderate awareness of complications. While they showed strong support

and collaboration, acceptance of the condition was comparatively lower.

Caregiving competency was generally high, particularly in medication

adherence, but weaker in managing physical activity. No significant cor-

relation was found between disease awareness and either disposition or

caregiving competency. However, disposition was significantly associated

with caregiving competency, with acknowledgment of role and support

and collaboration emerging as significant predictors.

Conclusion: Parental disposition and caregiving skills significantly

impact children’s health. Developing structured training programs is

crucial to enhance caregiving competency.

INITIAL STUDY ON TRUST COMPUTATION BASED ON TENSOR AND LIE ALGEBRA IN COMPLEX NETWORKS

Tran Dinh Que — 2026-09-09

Trust computation in complex networks is a crucial topic for systems

involving interactions among agents, such as recommendation platforms,

multi-agent systems, and social networks. This paper presents a novel

framework combining tensor-based feature modeling and Lie algebraic

dynamics to formally define and compute trust. We model measures

such as dispatch, familiarity, responsibility, and influence in tensor form

and propose a trust function based on weighted features. We then use Lie

algebra to encode the structural evolution of trust over time and derive

key mathematical properties. This approach enables robust and dynamic

trust inference in large-scale and evolving networks.

VALUE CHAIN ANALYSIS AND DESIGN OF A SUSTAINABLE CONSUMPTION LINKAGE MODEL FOR PHAN RANG GARLIC IN VINH HAI COMMUNE, KHANH HOA PROVINCE

Phan Phuoc Hien — 2026-09-09

Phan Rang garlic (Allium sativum L. var. Phan Rang) is an indige-

nous variety with the highest allicin content among domestic Vietnamese

garlic cultivars (1.2-1.8 mg/g fresh weight), cultivated in the arid ecologi-

cal zone of Vinh Hai commune, Khanh Hoa province. The study employs

a mixed-methods approach, surveying 100 respondents across four stake-

holder groups: 45 farming households, 10 distribution agents, 20 buyers

and 25 end consumers, combined with SPSS statistical analysis and Value

Chain Analysis (VCA). Findings confirm that Phan Rang garlic faces a

quality-market paradox: product quality index TC3 = 4.67/5 whereas

distribution linkage TC6 = 1.40/5 and price stability TC7 = 1.87/5;

farmers capture only 22-27% of the final retail price while post-harvest

loss rates reach 15-25%. Pearson correlation analysis shows r(TC6×TC7)

= +0.491, establishing distribution linkage as the most critical strategic

lever for improving price stability. The study proposes a five-tier inte-

grated linkage model combining digital raw material zone management,

brand identity, origin traceability and e-commerce, with a 10-indicator

KPI monitoring framework for 2026-2030.