OBJECTIVE: The objective of this research is to explore the applicability of machine learning and fully homomorphic encryption (FHE) in the private pathological assessment, with a focus on the inference phase of support vector machines (SVM) for the classification of confidential medical data.
METHODS: A framework is introduced that utilizes the Cheon-Kim-Kim-Song (CKKS) FHE scheme, facilitating the execution of SVM inference on encrypted datasets. This framework ensures the privacy of patient data and negates the necessity of decryption during the analytical process. Additionally, an efficient feature extraction technique is presented for the transformation of medical imagery into vectorial representations.
RESULTS: The system\'s evaluation across various datasets substantiates its practicality and efficacy. The proposed method delivers classification accuracy and performance on par with traditional, non-encrypted SVM inference, while upholding a 128-bit security level against established cryptographic attacks targeting the CKKS scheme. The secure inference process is executed within a temporal span of mere seconds.
CONCLUSIONS: The findings of this study underscore the viability of FHE in enhancing the security and efficiency of bioinformatics analyses, potentially benefiting fields such as cardiology, oncology, and medical imagery. The implications of this research are significant for the future of privacy-preserving machine learning, promoting progress in diagnostic procedures, tailored medical treatments, and clinical investigations.

BACKGROUND: Depression is very prevalent in middle-aged and older smokers. Therefore, we aimed to identify the risk of depression among middle-aged and older adults with frequent and infrequent nicotine use, as this is quite necessary for supporting their well-being.
METHODS: This study included a total of 10,821 participants, which were derived from the China Health and Retirement Longitudinal Study Wave 5, 2020 (CHARLS-5). Five machine learning (ML) algorithms were employed. Some metrics were used to evaluate the performance of models, including area under the receiver operating characteristic curve (AUC), positive predictive value (PPV), specificity, accuracy.
RESULTS: 10,821 participants (6472 males, 4349 females) had a mean age of 60.47 ± 8.98, with a score of 8.90 ± 6.53 on depression scale. For middle-aged and older adults with frequent nicotine use, random forest (RF) achieved the highest AUC value, PPV and specificity (0.75, 0.74 and 0.88, respectively). For the other group, support vector machines (SVM) showed the highest PPV (0.74), and relatively high accuracy and specificity (0.72 and 0.87, respectively). Feature importance analysis indicated that \"dissatisfaction with life\" was the most important variable of identifying the risk of depression in the SVM model, while \"attitude towards expected life span\" was the most important one in the RF model.
CONCLUSIONS: CHARLS-5 was collected during the COVID-19, so our results may be influenced by the pandemic.
CONCLUSIONS: This study indicated that certain ML models can ideally identify the risk of depression in middle-aged and older adults, which holds significant value for their health management.

The objective of the study was to develop a computational model with which predictions regarding the level of prevalence of mastitis in dairy sheep farms could be performed. Data for the construction of the model were obtained from a large Greece-wide field study with 111 farms. Unsupervised learning methodology was applied for clustering data into two clusters based on 18 variables (17 independent variables related to health management practices applied in farms, climatological data at the locations of the farms, and the level of prevalence of subclinical mastitis as the target value). The K-means tool showed the highest significance for the classification of farms into two clusters for the construction of the computational model: median (interquartile range) prevalence of subclinical mastitis among farms was 20.0% (interquartile range: 15.8%) and 30.0% (16.0%) (p = 0.002). Supervised learning tools were subsequently used to predict the level of prevalence of the infection: decision trees, k-NN, neural networks, and Support vector machines. For each of these, combinations of hyperparameters were employed; 83 models were produced, and 4150 assessments were made in total. A computational model obtained by means of Support vector machines (kernel: \'linear\', regularization parameter C = 3) was selected. Thereafter, the model was assessed through the results of the prevalence of subclinical mastitis in 373 records from sheep flocks unrelated to the ones employed for the selection of the model; the model was used for evaluation of the correct classification of the data in each of 373 sets, each of which included a test (prediction) subset with one record that referred to the farm under assessment. The median prevalence of the infection in farms classified by the model in each of the two categories was 10.4% (5.5%) and 36.3% (9.7%) (p < 0.0001). The overall accuracy of the model for the results presented by the K-means tool was 94.1%; for the estimation of the level of prevalence (<25.0%/≥25.0%) in the farms, it was 96.3%. The findings of this study indicate that machine learning algorithms can be usefully employed in predicting the level of subclinical mastitis in dairy sheep farms. This can facilitate setting up appropriate health management measures for interventions in the farms.

UNASSIGNED: Despite the growing body of evidence highlighting the individuality in movement techniques, predominant models of motor learning, particularly during the acquisition phase, continue to emphasise generalised, person-independent approaches. Biomechanical studies, coupled with machine learning approaches, have demonstrated the uniqueness of movement techniques exhibited by individuals. However, this evidence predominantly pertains to already stabilised movement techniques, particularly evident in cyclic daily activities such as walking, running, or cycling, as well as in expert-level sports movements. This study aims to evaluate the hypothesis of individuality in whole-body movements necessitating intricate coordination and strength among novice participants at the very beginning of an acquisition phase.
UNASSIGNED: In a within-subject design, sixteen highly active male participants (mean age: 23.1 ± 2.1 years), all absolute novices in the learning task (i.e., power snatch of Olympic weightlifting), participated in randomised snatch learning bouts. These bouts comprised 36 trials across various motor learning models: differential learning contextual interference (serial, sCIL; and blocked, bCIL), and repetitive learning. Kinematic and kinetic data were collected from three standardised snatch trials performed following each motor learning model bout. The time-continuous data were input to a linear Support Vector Machine (SVM). We conducted analyses on two classification tasks: participant and motor learning model.
UNASSIGNED: The Support Vector Machine classification revealed a notably superior participant classification compared to the motor learning model classification, with an averaged prediction accuracy of 78% (in average ≈35 out of 45 test trials across the folds) versus 27.3% (in average ≈9 out of 36 test trials across the folds). In specific fold and input combinations, accuracies of 91% versus 38% were respectively achieved.
UNASSIGNED: Methodically, the crucial role of selecting appropriate data pre-processing methods and identifying the optimal combinations of SVM data inputs is discussed in the context of future research. Our findings provide initial support for a dominance of individuality over motor learning models in movement techniques during the early phase of acquisition in Olympic weightlifting power snatch.

Tremor, defined as an \"involuntary, rhythmic, oscillatory movement of a body part\", is a key feature of many neurological conditions including Parkinson\'s disease and essential tremor. Clinical assessment continues to be performed by visual observation with quantification on clinical scales. Methodologies for objectively quantifying tremor are promising but remain non-standardized across centers. Our center performs full-body behavioral testing with 3D motion capture for clinical and research purposes in patients with Parkinson\'s disease, essential tremor, and other conditions. The objective of this study was to assess the ability of several candidate processing pipelines to identify the presence or absence of tremor in kinematic data from patients with confirmed movement disorders and compare them to expert ratings from movement disorders specialists. We curated a database of 2272 separate kinematic data recordings from our center, each of which was contemporaneously annotated as tremor present or absent by a movement physician. We compared the ability of six separate processing pipelines to recreate clinician ratings based on F1 score, in addition to accuracy, precision, and recall. The performance across algorithms was generally comparable. The average F1 score was 0.84±0.02 (mean ± SD; range 0.81-0.87). The second highest performing algorithm (cross-validated F1=0.87) was a hybrid that used engineered features adapted from an algorithm in longstanding clinical use with a modern Support Vector Machine classifier. Taken together, our results suggest the potential to update legacy clinical decision support systems to incorporate modern machine learning classifiers to create better-performing tools.

Tampered multimedia content is being increasingly used in a broad range of cybercrime activities. The spread of fake news, misinformation, digital kidnapping, and ransomware-related crimes are amongst the most recurrent crimes in which manipulated digital photos and videos are the perpetrating and disseminating medium. Criminal investigation has been challenged in applying machine learning techniques to automatically distinguish between fake and genuine seized photos and videos. Despite the pertinent need for manual validation, easy-to-use platforms for digital forensics are essential to automate and facilitate the detection of tampered content and to help criminal investigators with their work. This paper presents a machine learning Support Vector Machines (SVM) based method to distinguish between genuine and fake multimedia files, namely digital photos and videos, which may indicate the presence of deepfake content. The method was implemented in Python and integrated as new modules in the widely used digital forensics application Autopsy. The implemented approach extracts a set of simple features resulting from the application of a Discrete Fourier Transform (DFT) to digital photos and video frames. The model was evaluated with a large dataset of classified multimedia files containing both legitimate and fake photos and frames extracted from videos. Regarding deepfake detection in videos, the Celeb-DFv1 dataset was used, featuring 590 original videos collected from YouTube, and covering different subjects. The results obtained with the 5-fold cross-validation outperformed those SVM-based methods documented in the literature, by achieving an average F1-score of 99.53%, 79.55%, and 89.10%, respectively for photos, videos, and a mixture of both types of content. A benchmark with state-of-the-art methods was also done, by comparing the proposed SVM method with deep learning approaches, namely Convolutional Neural Networks (CNN). Despite CNN having outperformed the proposed DFT-SVM compound method, the competitiveness of the results attained by DFT-SVM and the substantially reduced processing time make it appropriate to be implemented and embedded into Autopsy modules, by predicting the level of fakeness calculated for each analyzed multimedia file.

The fight against bacterial antibiotic resistance must be given critical attention to avert the current and emerging crisis of treating bacterial infections due to the inefficacy of clinically relevant antibiotics. Intrinsic genetic mutations and transferrable antibiotic resistance genes (ARGs) are at the core of the development of antibiotic resistance. However, traditional alignment methods for detecting ARGs have limitations. Artificial intelligence (AI) methods and approaches can potentially augment the detection of ARGs and identify antibiotic targets and antagonistic bactericidal and bacteriostatic molecules that are or can be developed as antibiotics. This review delves into the literature regarding the various AI methods and approaches for identifying and annotating ARGs, highlighting their potential and limitations. Specifically, we discuss methods for (1) direct identification and classification of ARGs from genome DNA sequences, (2) direct identification and classification from plasmid sequences, and (3) identification of putative ARGs from feature selection.

Precision medicine is a framework for developing evidence-based medical recommendations that seeks to determine the optimal sequence of treatments tailored to all of the relevant patient-level characteristics which are observable. Because precision medicine relies on highly sensitive, patient-level data, ensuring the privacy of participants is of great importance. Dynamic treatment regimes (DTRs) provide one formalization of precision medicine in a longitudinal setting. Outcome-Weighted Learning (OWL) is a family of techniques for estimating optimal DTRs based on observational data. OWL techniques leverage support vector machine (SVM) classifiers in order to perform estimation. SVMs perform classification based on a set of influential points in the data known as support vectors. The classification rule produced by SVMs often requires direct access to the support vectors. Thus, releasing a treatment policy estimated with OWL requires the release of patient data for a subset of patients in the sample. As a result, the classification rules from SVMs constitute a severe privacy violation for those individuals whose data comprise the support vectors. This privacy violation is a major concern, particularly in light of the potentially highly sensitive medical data which are used in DTR estimation. Differential privacy has emerged as a mathematical framework for ensuring the privacy of individual-level data, with provable guarantees on the likelihood that individual characteristics can be determined by an adversary. We provide the first investigation of differential privacy in the context of DTRs and provide a differentially private OWL estimator, with theoretical results allowing us to quantify the cost of privacy in terms of the accuracy of the private estimators.

BACKGROUND: The early detection of Non-Melanoma Skin Cancer (NMSC) is essential to ensure patients receive the most effective treatment. Diagnostic screening tools for NMSC are crucial due to high confusion rates with other types of skin lesions, such as Actinic Keratosis. Nevertheless, current means of diagnosing and screening patients rely on either visual criteria, that are often conditioned by subjectivity and experience, or highly invasive, slow, and costly methods, such as histological diagnoses. From this, the objectives of the present study are to test if classification accuracies improve in the Near-Infrared region of the electromagnetic spectrum, as opposed to previous research in shorter wavelengths.
METHODS: This study utilizes near-infrared hyperspectral imaging, within the range of 900.6 and 1454.8 nm. Images were captured for a total of 125 patients, including 66 patients with Basal Cell Carcinoma, 42 with cutaneous Squamous Cell Carcinoma, and 17 with Actinic Keratosis, to differentiate between healthy and unhealthy skin lesions. A combination of hybrid convolutional neural networks (for feature extraction) and support vector machine algorithms (as a final activation layer) was employed for analysis. In addition, we test whether transfer learning is feasible from networks trained on shorter wavelengths of the electromagnetic spectrum.
RESULTS: The implemented method achieved a general accuracy of over 80 %, with some tasks reaching over 90 %. F1 scores were also found to generally be over the optimal threshold of 0.8. The best results were obtained when detecting Actinic Keratosis, however differentiation between the two types of malignant lesions was often noted to be more difficult. These results demonstrate the potential of near-infrared hyperspectral imaging combined with advanced machine learning techniques in distinguishing NMSC from other skin lesions. Transfer learning was unsuccessful in improving the training of these algorithms.
CONCLUSIONS: We have shown that the Near-Infrared region of the electromagnetic spectrum is highly useful for the identification and study of non-melanoma type skin lesions. While the results are promising, further research is required to develop more robust algorithms that can minimize the impact of noise in these datasets before clinical application is feasible.

UNASSIGNED: Spastic cerebral palsy, the most common pediatric-onset disabling condition with an estimated prevalence of 0.2% in children, is a complex condition characterized by stiff movement, muscle contractures, and abnormal gait that can diminish quality of life. Spastic CP accounts for approximately 83% of all CP cases and frequently co-occurs with other complex conditions, like epilepsy. An estimated 42% of spastic CP cases have co-occurring epilepsy. Unfortunately, CP is often difficult to diagnose. Although most children with CP are born with it or acquire it immediately after birth, many are not identified until after 19 months of age with CP diagnosis often not confirmed until 5 years of age. New bioinformatic approaches to identify CP earlier are needed. Recent studies indicate that altered DNA methylation patterns associated with CP may have diagnostic value. The potential confounding effects of co-occurrent epilepsy on these patterns are not known. We evaluated machine learning classification of CP patients with or without co-occurring epilepsy.
UNASSIGNED: Whole blood samples were collected from 30 study participants diagnosed with epilepsy (n=4), spastic CP (n=10), both (n=8), or neither (n=8). A novel Support-Vector-Machine learning algorithm was developed to identify methylation loci that have ability to classify CP from controls in the presence or absence of epilepsy. This algorithm was also employed to measure classification ability of identified methylation loci. After preprocessing of data, isolation of important methylation loci was performed in a binary comparison between CP and controls, as well as in a 4-way scheme, encapsulating epilepsy diagnoses. The classification ability was similarly assessed. CP Classification performance was evaluated with and without inclusion of epilepsy as a feature. Median F1 scores were 0.67 in 4-class comparison, and 1.0 in the binary classification, outperforming Linear-Discriminant-Analysis (0.57 and 0.86, respectively).
UNASSIGNED: This novel algorithm was able to classify study participants with spastic CP and/or epilepsy from controls with significant performance. The algorithm shows promise for rapid identification in methylation data of diagnostic methylation loci. In this model, Support Vector Machines outperformed Linear Discriminant Analysis in classification. In the evaluation of epigenetics-based diagnostics for CP, epilepsy may not be a significant confounding factor.