Claim Missing Document
Check
Articles

Found 8 Documents
Search

Frugal Innovation in Education: Designing and Evaluating Low-Bandwidth, Asynchronous Learning Systems for Remote Indonesian Schools Hesti Putri; Maya Enderson; Jasmila Tanjung; Matilda Munoz; Sarah Armalia; Jovanka Andina; Kevin Setiawan; Sudarto Sudarto; Khalil Jibran; Jasmine Alieva
Enigma in Education Vol. 3 No. 1 (2025): Enigma in Education
Publisher : Enigma Institute

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61996/edu.v3i1.95

Abstract

Background. Educational technology (EdTech) is often said to democratize learning in Indonesia. In remote island regions the promise falls short: poor connectivity and thin infrastructure keep the digital divide in place. Frugal innovation builds effective solutions from limited resources and may offer a context-aware response. Objective. We designed, implemented, and evaluated “Lentera”, a low-bandwidth, asynchronous, offline-first learning system for primary literacy and numeracy in remote Indonesian schools. Methods. Mixed-methods Design-Based Research came first. A school-level controlled evaluation followed over one academic year in six primary schools in the Maluku Islands (three intervention, three control; 302 students; 34 teachers). Data comprised literacy and numeracy pretests and posttests, System Usability Scale (SUS) surveys, usage logs, interviews with 18 teachers, and more than 80 hours of classroom observation. Students were nested in schools, so we fitted a two-level Hierarchical Linear Model (HLM). Results. Lentera was feasible to implement. Teachers received more than twice as many modules through offline peer-to-peer sharing as through internet downloads. After controlling for pretest scores, the intervention group scored higher in literacy (γ = 11.85, p < 0.001) and numeracy (γ = 12.91, p < 0.001). Mean SUS scores were 81.2 (teachers) and 75.8 (students). Interviews and observations described more student-centered, differentiated teaching and closer collaboration among teachers. Conclusion. A frugally designed, offline-first learning system was feasible. It was associated with higher literacy and numeracy in remote schools, in line with the aims of Kurikulum Merdeka. Replication with more schools is needed before the effects can be generalized.
Evaluating Generative AI as a Pedagogical Tool for Creative Problem-Solving in University Classrooms Hesti Putri; Khalil Jibran
Enigma in Education Vol. 3 No. 2 (2025): Enigma in Education
Publisher : Enigma Institute

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61996/edu.v3i2.117

Abstract

This study investigated the effectiveness of generative artificial intelligence (AI) as a pedagogical tool for enhancing creative problem-solving (CPS) skills and divergent thinking (DT) among undergraduate students. A quasi-experimental pretest-posttest control group design was employed with 120 education students at a private university in Palembang, Indonesia. Participants were assigned to an experimental group (n = 60) that engaged in a 12-week AI-assisted learning intervention within an Educational Psychology course, and a control group (n = 60) receiving conventional instruction. Creative problem-solving was measured using the Creative Problem-Solving Performance Inventory (CPSPI, α = 0.89), while divergent thinking was assessed through an adapted Torrance Tests of Creative Thinking (TTCT, ICC = 0.91). Results from mixed ANOVA revealed a significant interaction effect for CPS, F(1,118) = 89.34, p < .001, partial η² = 0.431. MANOVA confirmed significant multivariate differences across all outcome measures, Pillai’s V = 0.482, F(5,114) = 21.24, p < .001. Large effect sizes were observed for CPS (Hedges’ g = 1.66) and DT (Hedges’ g = 1.18). These findings suggest that structured integration of generative AI into university pedagogy can substantially improve students’ creative problem-solving and divergent thinking capacities.
Radiomics-Based Machine Learning for Automated Detection and Rupture-Risk Stratification of Cerebral Vascular Malformations: A Retrospective Cohort Study Hesti Putri; Nur Diana; Paula Magna Pablo-Rodriguez; Nadia Khoirina
Sriwijaya Journal of Radiology and Imaging Research Vol. 3 No. 2 (2025): Sriwijaya Journal of Radiology and Imaging Research
Publisher : Phlox Institute: Indonesian Medical Research Organization

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59345/sjrir.v3i2.291

Abstract

Introduction: Cerebral vascular malformations (CVMs) are the leading cause of spontaneous intracranial haemorrhage, yet their detection on CT/MR angiography is operator-dependent and existing machine-learning models derive almost exclusively from Caucasian or East Asian cohorts. We developed and internally validated a population-specific radiomics pipeline for CVM detection and rupture-risk stratification in a Southeast Asian population. Methods: In this retrospective diagnostic-and-prognostic cohort (STARD 2015; CLAIM) at a tertiary hospital in Palembang, Indonesia (2020–2025), 486 adults with diagnostic-quality CTA or TOF-MRA were analysed. After resampling and normalisation, 1,218 PyRadiomics features were reduced by LASSO and used to train Random Forest, SVM and XGBoost models (70/30 split). The reference standard was blinded consensus segmentation by two consultant neuroradiologists. Diagnostic accuracy (Wilson CI), AUC (DeLong), likelihood ratios, Cohen κ, McNemar test and a multivariable rupture model were computed. Results: CVM prevalence was 42.8% (208/486). XGBoost was the best detector (AUC 0.963 (95% CI 0.950–0.977); sensitivity 88.5 (95% CI 83.4–92.1)%; specificity 92.4 (95% CI 88.7–95.0)%; LR+ 11.71 (95% CI 7.74–17.72)), outperforming a single radiologist (ΔAUC 0.128, p<0.001; McNemar χ2=9.72, p=0.002 (discordant pairs b=73, c=39)). Inter-reader agreement was almost perfect (κ 0.845 (95% CI 0.797–0.893)). A radiomics-clinical model stratified rupture (AUC 0.846 (95% CI 0.792–0.901)), with lesion size (OR 4.26 (95% CI 2.60–6.98)) and hypertension (OR 2.46 (95% CI 1.18–5.14)) dominant and good calibration (χ2=14.87, df=8, p=0.062). Conclusion: A population-specific radiomics machine-learning pipeline achieved high diagnostic accuracy for CVM detection and clinically useful rupture-risk stratification, supporting operator-independent neurovascular triage in Southeast Asian settings. External validation is warranted.
Development and Validation of an Explainable Machine-Learning Model to Predict Cumulative Live Birth in PCOS-Associated Infertility Anies Fatmawati; Hesti Putri; Theresia Putri Sinaga
Sriwijaya Journal of Obstetrics and Gynecology Vol. 3 No. 2 (2025): Sriwijaya Journal of Obstetrics & Gynecology
Publisher : Phlox Institute: Indonesian Medical Research Organization

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59345/sjog.v3i2.282

Abstract

Introduction: Polycystic ovary syndrome (PCOS) is the leading cause of anovulatory infertility, yet predicting the cumulative live birth rate (CLBR) after in vitro fertilization (IVF) is difficult because of marked clinical heterogeneity. Conventional linear models discriminate modestly and most machine-learning (ML) tools remain uninterpretable. We aimed to develop and internally validate an explainable ML model for CLBR in women with PCOS. Methods: In a multicenter retrospective cohort at two tertiary reproductive-medicine centers in Indonesia (January 2018–December 2023), 1,245 women with PCOS (Rotterdam criteria) undergoing a first IVF/intracytoplasmic sperm injection cycle were analysed. Five algorithms (logistic regression, support vector machine, random forest, gradient boosting and eXtreme Gradient Boosting [XGBoost]) were trained (70%) and internally validated (30%) with 5-fold cross-validation. Discrimination used the area under the ROC curve (AUC) with DeLong confidence intervals (CI); SHapley Additive exPlanations (SHAP) quantified feature importance. Multivariable logistic regression, calibration and number-needed-to-treat (NNT) were also derived. Results: Overall CLBR was 54.6% (95% CI 51.8–57.4%). XGBoost performed best (AUC 0.82, 95% CI 0.79–0.85; accuracy 76.4%; sensitivity 78.2%; specificity 74.1%) and significantly exceeded logistic regression (AUC 0.72; ΔAUC 0.10, p<0.001). SHAP ranked top-quality embryo number, maternal age, anti-Müllerian hormone and oocyte yield as dominant predictors. Each additional top-quality embryo raised CLBR odds (adjusted OR 1.85, 95% CI 1.66–2.06, p<0.001); ≥3 versus <3 embryos yielded an NNT of 3.3. Conclusion: An explainable XGBoost model accurately predicts CLBR in PCOS-associated infertility and can support individualised counselling and cycle-management decisions. Prospective external validation is warranted before clinical deployment.
Machine learning identification of psychosocial and sociodemographic determinants of repeat adolescent pregnancy: a retrospective cohort study in an Indonesian metropolitan setting Rachmat Hidayat; Hesti Putri; Mahmood Abbas
Sriwijaya Journal of Obstetrics and Gynecology Vol. 4 No. 1 (2026): Sriwijaya Journal of Obstetrics & Gynecology
Publisher : Phlox Institute: Indonesian Medical Research Organization

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59345/sjog.v4i1.295

Abstract

Background: Repeat adolescent pregnancy compounds obstetric and socioeconomic risk, yet conventional models seldom capture the non-linear interplay of psychosocial and sociodemographic determinants, particularly in low- and middle-income registries where such variables are rarely recorded. Objective: To develop and compare interpretable machine-learning models to identify determinants of repeat adolescent pregnancy in an Indonesian metropolitan cohort. Methods: In this retrospective cohort study, electronic medical records (2018–2023) from a tertiary referral hospital and affiliated primary-care network in Palembang, Indonesia, were analysed for 1,245 adolescents (aged 10–19 years) with at least one prior pregnancy. The outcome was a second pregnancy before age 20. Five models (logistic regression, random forest, support vector machine, multilayer perceptron, and XGBoost) were trained with SMOTE and 5-fold cross-validation; discrimination was assessed by AUC-ROC and interpretability by SHAP. A multivariable logistic model provided adjusted odds ratios (aOR). Results: Repeat pregnancy occurred in 312/1,245 adolescents (25.06%; 95% CI 22.7–27.5%). XGBoost achieved the highest discrimination (AUC 0.89, 95% CI 0.86–0.92; F1 0.84). Independent determinants were non-use of postpartum LARC (aOR 8.86, 95% CI 6.00–13.07; p<0.001), low family support (aOR 3.82; p<0.001), education at most junior high (aOR 3.76; p<0.001), elevated EPDS (aOR 2.92, 95% CI 1.95–4.37; p<0.001), and age under 16 at first pregnancy (aOR 2.85; p<0.001). Postpartum LARC was strongly preventive (NNT approximately 3). Conclusion: Interpretable gradient boosting accurately stratified repeat adolescent pregnancy risk, and psychosocial determinants carried predictive weight comparable to contraceptive non-use. These findings support risk-stratified, bio-psycho-social postpartum care and targeted LARC counselling for adolescent mothers in Indonesia.
Integrating Kearifan Lokal (Local Wisdom) with Climate Adaptation Strategies: A Participatory Action Research on Enhancing Community Resilience and Achieving SDG 13 in Indonesia's Coastal Communities Jasmila Tanjung; Caelin Damayanti; Neva Dian Permana; Andi Fatihah Syahrir; Hesti Putri; Aman Suparman; Susi Diana
Indonesian Community Empowerment Journal Vol. 5 No. 2 (2025): Indonesian Community Empowerment Journal
Publisher : HM Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37275/icejournal.v5i2.45

Abstract

Coastal communities in Indonesia face existential threats from climate change. Conventional top-down adaptation strategies often fail due to a disconnect from local socio-ecological realities, overlooking a critical resource: traditional ecological knowledge, or kearifan lokal. This study investigates a knowledge co-production model that synergizes kearifan lokal with modern climate science to build community resilience. We employed a 24-month, mixed-methods Participatory Action Research (PAR) design in three highly exposed coastal villages in North Java, Indonesia. Ethical protocols, including Free, Prior, and Informed Consent (FPIC), were foundational. Qualitative data were gathered from in-depth interviews (n=30), focus groups (n=12), and ethnographic observation. Quantitative data came from a pre-test/post-test household survey (n=450) measuring a validated, multi-dimensional Community Resilience Index (CRI). Interventions were co-designed, blending traditional practices like the pranata mangsa (ethno-astronomical calendar) and the wana tirta (mangrove philosophy) with scientific recommendations. A linear mixed-effects model was used to analyze changes in CRI scores. The co-designed strategies led to a statistically significant increase in the mean CRI from a baseline of 2.8 (SD=0.65) to 4.2 (SD=0.48) post-intervention (p<0.001). Significant improvements were observed across all resilience dimensions, most notably in Economic Capital (+59.1%) and Adaptive Capacity & Governance (+51.7%). The revitalization of practices such as the restoration of 50 hectares of mangroves, guided by both wana tirta principles and scientific species selection, enhanced coastal protection and local livelihoods. In conclusion, the co-production of knowledge, facilitated through a PAR framework, is a potent mechanism for building effective, culturally embedded, and sustainable climate resilience. This model empowers communities as active agents in their adaptation journey and offers a scalable, evidence-based pathway for achieving SDG 13 in Indonesia and other climate-vulnerable nations.
Systemic Contagion or Digital Diversifier? A Dynamic Quantification of the Cryptocurrency Market's Evolving Role in Global Financial Risk Transmission Abdul Malik; Gayatri Putri; Hesti Putri; Ahmad Badruddin
Enigma in Economics Vol. 3 No. 2 (2025): Enigma in Economics
Publisher : Enigma Institute

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61996/economy.v3i2.99

Abstract

The proliferation of crypto-assets has raised critical questions about their impact on global financial stability. This study rigorously investigates the structural evolution of the cryptocurrency market's role within the global financial system, testing the hypothesis that it has transitioned from a peripheral, shock-absorbing entity into a systemically significant transmitter of financial risk. We employ a Time-Varying Parameter Vector Autoregression (TVP-VAR) model on daily data from January 1, 2017, to December 31, 2024, examining the dynamic connectedness between a bespoke, rebalanced cryptocurrency index (CRIX20) and key global financial indicators (S&P 500, MSCI World, VIX, DXY). The econometric framework utilizes a Bayesian estimation approach with standard priors, a 200-day rolling window, and a 10-day forecast horizon for Generalized Forecast Error Variance Decompositions (GFEVD). Methodological robustness is confirmed through structural break tests and sensitivity analysis of the forecast horizon. Our findings reveal a profound structural transformation. Prior to mid-2020, the cryptocurrency market was a consistent net receiver of financial spillovers. A structural break, formally identified in the third quarter of 2020, marks a definitive regime shift. Post-break, the crypto market has become a significant and persistent net transmitter of risk to the traditional financial system. The total connectedness index for the entire system shows a marked secular increase, with the crypto market's contribution to systemic risk growing substantially. Gross spillover analysis confirms this shift is driven by a dramatic increase in risk transmission from the crypto market to other assets. In conclusion, the cryptocurrency market can no longer be considered an isolated ecosystem; it is now an integral and potentially destabilizing component of the global financial architecture. The era of crypto-assets as reliable diversifiers has waned, replaced by a new reality where shocks originating within this market pose a credible threat to broader financial stability. These findings present urgent challenges for regulatory oversight, systemic risk monitoring, and portfolio management.
System Quality, Information Quality, and Technical Bur-den in Relation to Discontinuance Intention in GenAI-EdSent Digital Learning Applications: An Exploratory Rule-Based Review Analysis Guided by the DeLone–McLean Framework Hesti Putri; Matilda Munoz
Enigma in Education Vol. 4 No. 2 (2026): Enigma in Education
Publisher : Enigma Institute

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61996/edu.v4i2.137

Abstract

Background: Continued use of digital learning platforms is often explained by perceived experience quality. Public application store reviews, however, offer no individual-level retention data, so it is unclear how far stated intentions can stand in for actual retention. Objective: To determine how often educational application reviews state explicit continuance or discontinuance intentions, and how these relate to complaints about system quality, information quality, and technical burden. Methods: GenAI-EdSent was analyzed as a cross-sectional review corpus. Multi-stage filtering of 480,831 catalog rows from 22 educational applications left 204,294 unique English reviews (2021 to 13 August 2025). Lexical rules separated explicit discontinuance, explicit continuance, and unstated intentions before DeLone–McLean and Expectation-Confirmation coding, without human annotation. Exact conditional odds ratios (ORs), 95% exact confidence intervals (CIs), Fisher's exact tests, and Benjamini–Hochberg false discovery rate (q) adjustment were used. Results: Only 220 reviews (0.108%) stated an explicit intention: 180 discontinuance and 40 continuance. System quality issues appeared in 10/180 (5.6%) versus 1/40 (2.5%) reviews (OR = 2.29; 95% exact CI: 0.31–102.08; p = 0.694), information quality issues in 13/180 (7.2%) versus 1/40 (2.5%) (OR = 3.02; 95% CI: 0.43–132.28; p = 0.474), and technical burden in 2/180 versus 0/40, an unbounded OR (95% CI: 0.04–∞; p = 1.000; all q = 1.00). Multivariable regression was withheld because only 34 continuance comparators existed across three eligible applications. Conclusion: Large review catalogs do not guarantee enough explicit intention outcomes for adjusted modeling. The estimates are imprecise crude associations, not measures of retention.