Science and Engineering Connect

Solid-state Fermentation of Palm Kernel Cake Using Two Bacillus Strains in Co-culture for Probiotics Production

Mon, 29 Jun 2026 00:00:00 +0700

Background and Objectives: Bacillus velezensis and Bacillus subtilis are bacteria commonly used as probiotic microorganisms in animal feed formulations due to their effectiveness in controlling and inhibiting pathogenic microbes. Additionally, they can produce beneficial enzymes that aid animal digestion, including proteases and amylases. Solid-state fermentation (SSF) is a widely adopted process for probiotics production and is also known to enhance the nutritional value of animal feed. However, previous studies have shown that using a single microbial strain in SSF often results in incomplete utilization of nutrients from raw materials at the end of the fermentation process. Therefore, two probiotic strains, B. velezensis and B. subtilis, were selected to enhance the efficiency of nutrient utilization from a raw material. This dual-strain approach is expected to improve probiotics production efficiency from agro-industrial by-products, specifically palm kernel cake, which is a residue from the palm oil production industry.

Methodology: This study employed SSF with B. velezensis and B. subtilis at the ratios of 1:1, 1:2, and 2:1 with palm kernel cake as the fermentation substrate. Fermentation was carried out over a period of 7 days at 37°C. Samples were collected on days 0, 1, 3, 5, and 7 for the analysis. Changes of various parameters were monitored throughout the fermentation period, including total viable count, soluble protein content, protease activity, amylase activity, reducing sugar content, moisture content and pH. The objective was to determine the optimal condition for probiotics production.

Main Results: The results of the study on the optimal ratio for fermenting palm kernel cake with B. velezensis and B. subtilis at ratios of 1:1, 1:2 and 2:1 reveal that the total viable microbial count reached its peak on day 3 for all tested ratios. Similarly, the protease and amylase enzyme activities in the fermented palm kernel cake reached their highest levels on day 3, corresponding with the peak in total viable microbial count. The samples with the ratios of 1:1 and 2:1 exhibited higher protease activity compared to the one with the ratio of 1:2; the 1:1 ratio showed the highest amylase activity among all the treatments.

Conclusions: Utilization of palm kernel cake as a substrate for SSF to produce probiotic microorganisms and to enhance nutritional value was investigated using two probiotic strains, B. velezensis and B. subtilis. A ratio of 1:1 was identified as the optimal condition at the laboratory scale.

Practical Application: The present findings can serve as a basis for scaling up the process to pilot-scale production and can be further developed into a commercial application.

Fine-tuning Fully Connected Layer Architecture in Convolutional Neural Networks for Tuberculosis Classification of Chest Radiographs

Pongsathorn Chedsom — Mon, 29 Jun 2026 00:00:00 +0700

Background and Objectives: Tuberculosis remains a leading cause of worldwide mortality, accounting for approximately 1.5 million deaths annually. Rapid diagnosis is essential for increasing accessibility to timely treatment. Although Convolutional Neural Network (CNN) technologies exhibit high performance, large-scale models often encounter over-parameterization, which may be unsuitable in resource-limited settings. The present research presents an experimental study to fine-tune 15 variations of Fully Connected (FC) layer architectures to identify a classification head that achieves an optimal balance between resource efficiency (lightweight) and maximum performance (effectiveness).

Methodology: The models were developed by pairing 15 popular feature extraction architectures (backbones) with 15 newly designed FC layer patterns, resulting in a total of 225 model combinations. A balanced dataset comprising 6,388 chest X-ray images (3,194 Tuberculosis and 3,194 Normal) was utilized to mitigate model bias. The data was partitioned into a training set (70%, 4,471 images), a validation set (20%, 1,278 images), and a test set (10%, 639 images). Performance was evaluated using 5-fold cross-validation to assess stability. Hyperparameters were set as follows: Adam optimizer (LR = 0.001), 15 epochs, and a batch size of 128.

Main Results: VGG16 architecture combined with "FC Pattern 11" achieved the highest performance, yielding an accuracy of 98.58%, a mean cross-validation accuracy of 97.52% (SD = 0.01), and a loss of 0.15. Clinical evaluation metrics demonstrated a sensitivity of 99.05% and a specificity of 98.11%. Pattern 11 employs a sequential node reduction strategy (384 -> 192 -> 96) integrated with Batch Normalization and L2 Regularization, which effectively improves internal stability and mitigates overfitting. Furthermore, this model maintains a lightweight structure with a depth of 16 layers, effectively reducing computational load and preventing over-parameterization more efficiently than deeper, more complex architectures.

Conclusions: Fine-tuning fully connected layers can significantly enhance standard architectures to a level suitable for Clinical Decision Support Systems (CDSS). The developed model is technically appropriate for preliminary screening, particularly due to its high sensitivity (99.05%), which minimizes the risk of missing pathological cases during initial triage.

Practical Application: The proposed model can be implemented as a preliminary screening tool to effectively reduce the workload of radiologists by filtering normal cases from suspected ones. The approach could help accelerate the diagnostic process, particularly in hospitals with limited resources and specialists, ensuring that patients receive timely and effective treatment.

From the Voice of Thai User: Exploring the Influence of Battery Electric Vehicles’ User Experience and Its Antecedents on Willingness to Recommend

Pannathadh Chomchark, Rawich Wongsawad, Pichit Ngamjarussrivichai — Mon, 29 Jun 2026 00:00:00 +0700

Background and Objectives: Between 2022 and 2025, the number of battery electric vehicle (BEV) users in Thailand significantly increased, with the average annual growth rate of BEV registrations as high as 524.2%. The Thai government has continuously supported the expansion of the electric vehicle industry through various policies, including encouraging both domestic and foreign investments in establishing local manufacturing bases, providing subsidies to support the purchase of electric vehicles, and implementing reductions in import duties and excise taxes. These measures aim to encourage Thai consumers to transition toward greater adoption of green energy in the transportation sector. The present study thus aimed to investigate the influence of battery electric vehicle user experience (BEVUX) and BEVs’ quality (BEVQUAL) affecting willingness to recommend.

Methodology: The study employed a quantitative research approach. Data were collected through an online questionnaire administered to BEV users in Thailand, aged 20 years and above, with a total sample size of 845 respondents. The statistical techniques used for data analysis included frequency and percentage, while structural equation modeling (SEM) was employed to analyze the causal relationships among the variables. A quantitative research design utilizing AMOS, which is particularly well-suited for analyzing complex structural models with multiples constructs, was used.

Main Results: 1) BEVQUAL exhibited a direct positive influence on BEVUX (β= 0.94, p < .001); 2) BEVUX exhibited a direct positive influence on continuance use intention (β= 0.76, p < .001); 3) Continuance use intention possessed a direct positive influence on willingness to recommend (β= 0.54, p < .001); 4) BEVUX exhibited a direct positive influence on willingness to recommend (β= 0.27, p < .01); 5) BEVQUAL exhibited a positive indirect influence on continuance use intention, mediated through BEVUX (β= 0.71, p < .001); 6) BEVQUAL had a positive indirect influence on willingness to recommend, mediated through BEVUX and continuance use intention (β= 0.65, p < .001); and 7) BEVUX possessed a positive indirect influence on willingness to recommend, mediated through continuance use intention (β= 0.41, p < .001).

Conclusions: BEVQUAL, BEVUX, and continuance use intention exerted both direct and indirect influences on willingness to recommend. The structural model demonstrated consistency and good fit with the empirical data.

Practical Application: The findings of the study can be utilized as useful information for manufacturers and distributors of BEV to further improve the research and development of electric vehicle technologies, strengthen consumer confidence in after-sales services, and deliver positive user experience. As a result, users may be more willing to confidently and positively share their experiences of using BEV with others through word-of-mouth communication.