Communication in Biomathematical Sciences

Optimal Land Site Selection for Organic Sugar Palm Plantations Using Spherical Picture Fuzzy Hamacher Aggregation Operators

2025-12-11T14:10:53+07:00

This study enhances the analytical capabilities of Picture Fuzzy Sets (PFSs) and Circular Intuitionistic Fuzzy Sets by introducing Spherical Picture Fuzzy Sets (SPFSs), which incorporate neutral membership and radius-based parameters for more comprehensive uncertainty modeling. The research specifically addresses computational limitations in conventional PFS evaluation methods through the development of Hamacher-based aggregation mechanisms within SPFS environments.
Novel Spherical Picture Fuzzy Hamacher aggregation operators, including both arithmetic and geometric variants, are formulated to support advanced Multi-Attribute Decision Making (MADM) models. The theoretical properties of these operators are established, and an SPFS ranking formula is integrated to evaluate alternatives. A numerical study involving agricultural land site selection for organic sugar palm plantation development is conducted to demonstrate computational feasibility.
The numerical examples confirm that the proposed Hamacher aggregation operators effectively capture complex uncertainty, hesitancy, and multi-dimensional evaluation criteria. The SPFS-based MADM framework successfully evaluates alternative land sites using factors such as soil fertility, water availability, climatic conditions, and infrastructure accessibility. The results validate the mathematical soundness and operational effectiveness of the proposed operators.
This work contributes a new class of Hamacher aggregation operators for SPFSs, enriching the theoretical foundation of fuzzy decision-making. The integration of arithmetic and geometric Hamacher operations with SPFS structures provides enhanced flexibility and precision in modeling sophisticated multi-criteria problems characterized by high uncertainty. The study demonstrates practical applicability through a real-world agricultural land selection scenario.

A Mathematical Model for the Co-infection of Malaria and Typhoid using Optimal Control Set-up Strategy

2025-12-10T23:13:20+07:00

Malaria and typhoid fever remain major public health challenges in sub-Saharan Africa, with frequent co-infection cases due to overlapping risk factors such as poor sanitation and vector presence. This study proposes a deterministic compartmental model using a system of nonlinear differential equations to study the dynamics of malaria-typhoid co-infection. The model incorporates optimal control strategies to evaluate the effectiveness of treatment and prevention measures. Analytical and numerical techniques will be used to study the behavior of the model, visualize optimal control trajectories and to design cost-effective control strategies to minimize both disease prevalence and control costs.

Biogeochemical model simulation of GHG emissions and yield: Sustaining dryland winter wheat systems with one-time composted cattle manure application and annual cover cropping

2025-12-06T21:21:34+07:00

Dryland winter wheat-fallow systems in the U.S. northern High Plains face persistent soil fertility and climate-related constraints that limit yields and long-term sustainability. This study monitored greenhouse gas (GHG) emissions and wheat yield following a one-time compost application (0, 15, 30, and 45 Mg ha⁻¹ ) in 2015 combined with annual fallow-phase cover cropping (2:1 Austrian winter pea and oats). The DAYCENT biogeochemical model was used to evaluate climate-smart management and determine the optimal timing for compost re-application from 2015 to 2050. Model validation with field measurements showed strong agreement between observed and simulated carbon dioxide (CO₂, R² = 0.62), methane (CH₄, R² = 0.64), nitrous oxide (N₂O, R² = 0.52), and grain yield (R² = 0.93). Simulations indicated that CO₂ and N₂O emissions peaked at 923 mg C m⁻² hr⁻¹ and 78 µg N m⁻² hr⁻¹ approximately 7 years after compost application, remaining elevated (923-1030 mg C m⁻² hr⁻¹ and 57-78 µg N m⁻² hr⁻¹) until 2031 and 2033 before declining. CH₄ remained a consistent sink, with 99% greater uptake in compost- and cover-crop-amended soils. Wheat yield declined slightly in the first 3-5 years, increased until year 15 (2031), then gradually decreased. Overall, a single 45 Mg ha⁻¹ compost application with annual cover cropping can enhance soil function, productivity, and GHG outcomes for 10-15 years, after which re-application is required to sustain benefits.

Ensemble Learning to Predict Hospital Stay Duration from Diagnostic and Drug Consumption Information

2025-11-28T09:55:27+07:00

Predicting the length of patient hospitalisation is an important challenge in hospital operational man
agement. Accurate predictive models are essential for efficient resource allocation and cost control. Recent
advances in machine learning, particularly with the use of complex data from electronic health records (EHR)
in specific regions such as Aceh, Indonesia, have opened up opportunities to develop more accurate and
reliable predictive tools. This study aims to develop and evaluate highly effective machine learning models
for predicting patient length of stay using demographic, diagnosis, and medication data from a hospital in
Aceh. A two-level stacking model was implemented, using LightGBM, XGBoost, CatBoost, Random Forest,
and AdaBoost as base models. All base models were optimised using Bayesian optimisation (50 iterations).
The final ensemble prediction was generated by an XGBoost meta-model trained on out-of-bag predictions
(five-fold cross-validation). Model performance was evaluated using RMSE, MAE, and R2, with five-fold
cross-validation to ensure generalisation. In addition, SHAP analysis was used to ensure the interpretability
of the ensemble model. The stacked ensemble model significantly outperformed all individual base models.
It achieved an average R2 of 0.88, RMSE of 1.73, and MAE of 1.11. These results represent a significant
9% improvement in R2 compared to the standalone LightGBM model (R2 = 0.79), demonstrating its superior
ability to capture complex nonlinear relationships and reduce prediction variance. The developed Stacked
Ensemble Model proved to be an accurate and interpretable tool for predicting length of stay (LOS), which
has important implications for clinical operational management and resource planning. Further research should
focus on validating this model using larger, multi-institutional datasets to ensure its broad applicability, as well
as exploring richer clinical features to improve its predictive power.