*Editorial* **Enhancing Health and Public Health through Machine Learning: Decision Support for Smarter Choices**

**Pedro Miguel Rodrigues 1,\*, João Paulo Madeiro <sup>2</sup> and João Alexandre Lobo Marques <sup>3</sup>**


In recent years, the integration of Machine Learning (ML) techniques in the field of healthcare and public health has emerged as a powerful tool for improving decision-making processes. The ability of ML algorithms to analyze vast amounts of data, identify patterns, and generate actionable insights has opened new avenues for enhancing various aspects of healthcare delivery and public health initiatives. This Special Issue (SI) explores the applications of ML in health and public health decision support systems, highlighting their potential benefits and challenges, mainly in the following areas:

	- Mirniaharikandehei et al. [5] explore the feasibility of using a modified deep learning (DL) method for automatically segmenting disease-infected regions and predicting disease severity in computed tomography (CT) images. A dataset from 20 COVID-19 patients has been used, incorporating manually annotated lung and infection masks. An ensemble DL model was trained, combining five customized residual attention U-Net models for disease-infected region segmentation and a Feature Pyramid Network model for disease severity stage prediction. The analysis reveals >90% agreement in disease severity classification between the DL model and radiologists for 45 testing cases.
	- Chen et al. [6] explore a noninvasive, cost-effective tool to assess the risk of subclinical renal damage (SRD) in asymptomatic individuals. Using ML algorithms, a risk assessment score model was established based on systolic blood pressure, diastolic blood pressure, and body mass index. The model demonstrated excellent classification ability, with an AUC value of 0.778 for SRD estimation and 0.729 for 4-year SRD risk prediction.
	- Zhang et al. [7] investigate the effects of atherosclerotic intracranial internal carotid artery stenosis (IICAS) on extracranial internal carotid artery (ICA) flow velocity waveforms to identify sensitive hemodynamic indices for IICAS diagnoses. Hemodynamic indices, including peak systolic velocity (PSV), enddiastolic velocity (EDV), resistive index (RI), and the first harmonic ratio (FHR), were analyzed in simulations with and without IICAS. In a case-control study

**Citation:** Rodrigues, P.M.; Madeiro, J.P.; Marques, J.A.L. Enhancing Health and Public Health through Machine Learning: Decision Support for Smarter Choices. *Bioengineering* **2023**, *10*, 792. https://doi.org/10.3390/ bioengineering10070792

Received: 1 June 2023 Accepted: 29 June 2023 Published: 2 July 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

with patients having mild-to-moderate IICAS, statistical analyses revealed that the average PSV, EDV, and RI were lower in the stenosis group compared to the control group, but without significant differences (*p* > 0.05), except for the PSV of the right ICA (*p* = 0.011). However, the FHR showed a significantly higher value in the stenosis group compared to the control group (*p* < 0.001), indicating its potential as a superior diagnostic index for early IICAS detection using carotid Doppler ultrasound methods.

	- Kim et al. [12] used transfer transformers to identify drug–drug and chemical– protein interactions. They utilized the DDI Extraction-2013 Shared Task and BioCreative ChemProt datasets for extracting drug-related interactions. Two models were proposed: BERTGAT, incorporating a graph attention network for sentence structure, and T5slim\_dec, adapting T5's generation task for relation classification. T5slim\_dec achieved remarkable performance with 91.15% accuracy on the DDI dataset and 94.29% accuracy for the CPR class group in ChemProt. However, BERTGAT did not significantly improve relation extraction. This highlights the language understanding capability of transformer-based

approaches, which can comprehend language effectively without relying on additional structural information.

	- Rodrigues et al. [15] introduced a hybrid method combining pre-trained CNN keras models and classical ML models to visually discriminate different bacterial colonies based on their morphology on culture media. The system achieved high accuracy rates: 92% for Pseudomonas aeruginosa vs. Staphylococcus aureus, 91% for Escherichia coli vs. Staphylococcus aureus, and 84% for Escherichia coli vs. Pseudomonas aeruginosa.
	- Promsri et al. [17] studied the relationship between walking stability and fall risk markers in older adults. Three-dimensional lower-limb kinematic data from 43 healthy individuals were analyzed using principal component analysis (PCA) to extract principal movements (PMs) representing different components of walking. The largest Lyapunov exponent (LyE) was applied to the PMs as a measure of stability. Fall risk was assessed using the Short Physical Performance Battery (SPPB) and the Gait Subscale of Performance-Oriented Mobility Assessment (POMA-G). Results indicated a negative correlation (*p* ≤ 0.009) between SPPB and POMA-G scores and LyE in specific PMs, suggesting that increased walking instability is associated with higher fall risk.
	- Gupta et al. study [18] aimed to detect and address stress, which is a significant factor affecting mental health and overall well-being. In this study, a novel approach utilizing audio-visual data processing is proposed to detect human mental stress. By employing the cascaded RNN-LSTM strategy, the study achieved a high accuracy of 91% in classifying emotions and distinguishing between stressed and unstressed states using the RAVDESS dataset.
	- da Silva et al. [20] proposed a methodology to analyze the performance of measurement systems during the design phase using the Monte Carlo method. The methodology was applied to a simulated ECG, estimating a measurement uncertainty of 3.54% with 95% confidence. The analysis revealed that the preamplifier module had a greater impact on the measurement results compared to the final stage module, suggesting that interventions in the preamplifier module would yield more significant improvements.

To conclude, ML has revolutionized health and public health decision support systems by enabling data-driven insights and informed decision-making. By harnessing the power of ML algorithms, healthcare professionals and public health authorities can improve disease diagnosis and prognosis, personalize treatment strategies, detect outbreaks, analyze health behaviors, and optimize resource allocation. As technology continues to advance, the integration of ML in health and public health applications will play an increasingly significant role in transforming healthcare delivery and improving population health outcomes.

**Author Contributions:** Conceptualization, P.M.R.; methodology, P.M.R.; validation, P.M.R., J.P.M. and J.A.L.M.; investigation, P.M.R.; writing—original draft preparation, P.M.R.; writing—review and editing, P.M.R., J.P.M. and J.A.L.M. All authors have read and agreed to the published version of the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
