-
Military Medicine Jun 2024Several challenges face the U.S. Marine Corps (USMC) and other services in their efforts to design recruit training to augment warfighter mobility and resilience in both...
Unsupervised Machine Learning in Countermovement Jump and Isometric Mid-Thigh Pull Performance Produces Distinct Combat and Physical Fitness Clusters in Male and Female U.S. Marine Corps Recruits.
INTRODUCTION
Several challenges face the U.S. Marine Corps (USMC) and other services in their efforts to design recruit training to augment warfighter mobility and resilience in both male and female recruits as part of an integrated model. Strength and power underpin many of the physical competencies required to meet the occupational demands one might face in military. As the military considers adopting force plate technology to assess indices of strength and power, an opportunity presents itself for the use of machine learning on large datasets to deduce the relevance of variables related to performance and injury risk. The primary aim of this study was to determine whether cluster analysis on baseline strength and power data derived from countermovement jump (CMJ) and isometric mid-thigh pull (IMTP) adequately partitions men and women entering recruit training into distinct performance clusters. The secondary aim of this study is then to assess the between-cluster frequencies of musculoskeletal injury (MSKI).
MATERIALS AND METHODS
Five hundred and sixty-five males (n = 386) and females (n = 179) at the Marine Corps Recruit Depots located at Parris Island and San Diego were enrolled in the study. Recruits performed CMJ and IMTP tests at the onset of training. Injury data were collected via medical chart review. Combat fitness test (CFT) and physical fitness test (PFT) results were provided to the study team by the USMC. A k-means cluster analysis was performed on CMJ relative peak power, IMTP relative peak force, and dynamic strength index. Independent sample t-tests and Cohen's d effect sizes assessed between-cluster differences in CFT and PFT performance. Differences in cumulative incidence of lower extremity %MSKIs were analyzed using Fisher's exact test. Relative risk and 95% confidence intervals (CIs) were also calculated.
RESULTS
The overall effects of cluster designation on CMJ and IMTP outcomes ranged from moderate (relative peak power: d = -0.68, 95% CI, -0.85 to -0.51) to large (relative peak force: d = -1.69, 95% CI, -1.88 to -1.49; dynamic strength index: d = 1.20, 95% CI, 1.02-1.38), indicating acceptable k-means cluster partitioning. Independent sample t-tests revealed that both men and women in cluster 2 (C2) significantly outperformed those in cluster 1 (C1) in all events of the CFT and PFT (P < .05). The overall and within-gender effect of cluster designation on both CFT and PFT performance ranged from small (d > 0.2) to moderate (d > 0.5). Men in C2, the high-performing cluster, demonstrated a significantly lower incidence of ankle MSKI (P = .04, RR = 0.2, 95% CI, 0.1-1.0). No other between-cluster differences in MSKI were statistically significant.
CONCLUSIONS
Our results indicate that strength and power metrics derived from force plate tests effectively partition USMC male and female recruits into distinct performance clusters with relevance to tactical and physical fitness using k-means clustering. These data support the potential for expanded use of force plates in assessing readiness in a cohort of men and women entering USMC recruit training. The ability to pre-emptively identify high and low performers in the CFT and PFT can aid in leadership developing frameworks for tailoring training to enhance combat and physical fitness with benchmark values of strength and power.
Topics: Humans; Female; Male; Military Personnel; Physical Fitness; Adult; Unsupervised Machine Learning; Cluster Analysis; Muscle Strength; Exercise Test; United States; Adolescent; Thigh
PubMed: 38920035
DOI: 10.1093/milmed/usad371 -
Minerva Urology and Nephrology Jun 2024Artificial intelligence and machine learning are the new frontier in urology; they can assist the diagnostic work-up and in prognostication bring superior to the...
INTRODUCTION
Artificial intelligence and machine learning are the new frontier in urology; they can assist the diagnostic work-up and in prognostication bring superior to the existing nomograms. Infectious events and in particular the septic risk, are one of the most common and in some cases life threatening complication in patients with urolithiasis. We performed a scoping review to provide an overview of the current application of AI in prediction the infectious complications in patients affected by urolithiasis.
EVIDENCE ACQUISITION
A systematic scoping review of the literature was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses for Scoping Reviews (PRISMA-ScR) guidelines by screening Medline, PubMed, and Embase to detect pertinent studies.
EVIDENCE SYNTHESIS
A total of 467 articles were found, of which nine met the inclusion criteria and were considered. All studies are retrospective and published between 2021 and 2023. Only two studies performed an external validation of the described models. The main event considered is urosepsis in four articles, urinary tract infection in two articles and diagnosis of infection stones in three articles. Different AI models were trained, each of which exploited several types and numbers of variables. All studies reveal good performance. Random forest and artificial neural networks seem to have higher AUC, specificity and sensibility and perform better than the traditional statistical analysis.
CONCLUSIONS
Further prospective and multi-institutional studies with external validation are needed to better clarify which variables and AI models should be integrated in our clinical practice to predict infectious events.
Topics: Humans; Urolithiasis; Artificial Intelligence; Urinary Tract Infections; Risk Assessment; Sepsis; Machine Learning
PubMed: 38920010
DOI: 10.23736/S2724-6051.24.05686-6 -
Frontiers in Surgery 2024Lymph node (LN) status is a vital prognostic factor for patients. However, there has been limited focus on predicting the prognosis of patients with late-onset gastric...
INTRODUCTION
Lymph node (LN) status is a vital prognostic factor for patients. However, there has been limited focus on predicting the prognosis of patients with late-onset gastric cancer (LOGC). This study aimed to investigate the predictive potential of the log odds of positive lymph nodes (LODDS), lymph node ratio (LNR), and pN stage in assessing the prognosis of patients diagnosed with LOGC.
METHODS
The LOGC data were obtained from the Surveillance, Epidemiology, and End Results database. This study evaluated and compared the predictive performance of three LN staging systems. Univariate and multivariate Cox regression analyses were carried out to identify prognostic factors for overall survival (OS). Three machine learning methods, namely, LASSO, XGBoost, and RF analyses, were subsequently used to identify the optimal LN staging system. A nomogram was built to predict the prognosis of patients with LOGC. The efficacy of the model was demonstrated through receiver operating characteristic (ROC) curve analysis and decision curve analysis.
RESULTS
A total of 4,743 patients with >16 removed lymph nodes were ultimately included in this investigation. Three LN staging systems demonstrated significant performance in predicting survival outcomes ( < 0.001). The LNR exhibited the most important prognostic ability, as evidenced by the use of three machine learning methods. Utilizing independent factors derived from multivariate Cox regression analysis, a nomogram for OS was constructed.
DISCUSSION
The calibration, C-index, and AUC revealed their excellent predictive performance. The LNR demonstrated a more powerful performance than other LN staging methods in LOGC patients after surgery. Our novel nomogram exhibited superior clinical feasibility and may assist in patient clinical decision-making.
PubMed: 38919979
DOI: 10.3389/fsurg.2024.1376702 -
Frontiers in Neurology 2024The aim of this study is to develop a predictive model utilizing deep learning and machine learning techniques that will inform clinical decision-making by predicting...
BACKGROUND
The aim of this study is to develop a predictive model utilizing deep learning and machine learning techniques that will inform clinical decision-making by predicting the 1-year postoperative recovery of patients with lumbar disk herniation.
METHODS
The clinical data of 470 inpatients who underwent tubular microdiscectomy (TMD) between January 2018 and January 2021 were retrospectively analyzed as variables. The dataset was randomly divided into a training set ( = 329) and a test set ( = 141) using a 10-fold cross-validation technique. Various deep learning and machine learning algorithms including Random Forests, Extreme Gradient Boosting, Support Vector Machines, Extra Trees, K-Nearest Neighbors, Logistic Regression, Light Gradient Boosting Machine, and MLP (Artificial Neural Networks) were employed to develop predictive models for the recovery of patients with lumbar disk herniation 1 year after surgery. The cure rate score of lumbar JOA score 1 year after TMD was used as an outcome indicator. The primary evaluation metric was the area under the receiver operating characteristic curve (AUC), with additional measures including decision curve analysis (DCA), accuracy, sensitivity, specificity, and others.
RESULTS
The heat map of the correlation matrix revealed low inter-feature correlation. The predictive model employing both machine learning and deep learning algorithms was constructed using 15 variables after feature engineering. Among the eight algorithms utilized, the MLP algorithm demonstrated the best performance.
CONCLUSION
Our study findings demonstrate that the MLP algorithm provides superior predictive performance for the recovery of patients with lumbar disk herniation 1 year after surgery.
PubMed: 38919973
DOI: 10.3389/fneur.2024.1255780 -
Frontiers in Genetics 2024Moyamoya disease (MMD) is a chronic cerebrovascular disease that can lead to ischemia and hemorrhagic stroke. The relationship between oxidative phosphorylation...
Moyamoya disease (MMD) is a chronic cerebrovascular disease that can lead to ischemia and hemorrhagic stroke. The relationship between oxidative phosphorylation (OXPHOS) and MMD pathogenesis remains unknown. The gene expression data of 60 participants were acquired from three Gene Expression Omnibus (GEO) datasets, including 36 and 24 in the MMD and control groups. Differentially expressed genes (DEGs) between MMD patients MMD and control groups were identified. Machine learning was used to select the key OXPHOS-related genes associated with MMD from the intersection of DEGs and OXPHOS-related gene sets. Gene ontology (GO), Kyoto encyclopedia of genes and genomes (KEGG), gene set enrichment analysis (GSEA), Immune infiltration and microenvironments analysis were used to analyze the function of key genes. Machine learning selected four key OXPHOS-related genes associated with MMD: , , and ( was upregulated and the other three were downregulated). Functional enrichment analysis showed that these genes were mainly enriched in the Notch signaling pathway, GAP junction, and RNA degradation, which are related to several biological processes, including angiogenesis, proliferation of vascular smooth muscle and endothelial cells, and cytoskeleton regulation. Immune analysis revealed immune infiltration and microenvironment in these MMD samples and their relationships with four key OXPHOS-related genes. APC co-inhibition ( = 0.032), HLA ( = 0.001), MHC I ( = 0.013), T cellco- inhibition ( = 0.032) and Type I IFN responses ( < 0.001) were significantly higher in the MMD groups than those in the control groups. The positively correlated with APC co-inhibition and T cell-co-inhibition. The negatively correlated with Type I IFN response. The negatively correlated with APC co-inhibition and Type I IFN response. The positively correlated with HLA, MHC I and Type I IFN responses. This study provides a comprehensive understanding of the role of OXPHOS in MMD and will contribute to the development of new treatment methods and exploration of MMD pathogenesis.
PubMed: 38919950
DOI: 10.3389/fgene.2024.1417329 -
Frontiers in Molecular Biosciences 2024This study bridges traditional remedies and modern pharmacology by exploring the synergy between natural compounds and Ceritinib in treating Non-Small Cell Lung Cancer...
This study bridges traditional remedies and modern pharmacology by exploring the synergy between natural compounds and Ceritinib in treating Non-Small Cell Lung Cancer (NSCLC), aiming to enhance efficacy and reduce toxicities. Using a combined approach of computational analysis, machine learning, and experimental procedures, we identified and analyzed PD173074, Isoquercitrin, and Rhapontin as potential inhibitors of fibroblast growth factor receptor 3 (FGFR3). Machine learning algorithms guided the initial selection, followed by Quantitative Structure-Activity Relationship (QSAR) modeling and molecular dynamics simulations to evaluate the interaction dynamics and stability of Rhapontin. Physicochemical assessments further verified its drug-like properties and specificity. Our experiments demonstrate that Rhapontin, when combined with Ceritinib, significantly suppresses tumor activity in NSCLC while sparing healthy cells. The molecular simulations and physicochemical evaluations confirm Rhapontin's stability and favorable interaction with FGFR3, highlighting its potential as an effective adjunct in NSCLC therapy. The integration of natural compounds with established cancer therapies offers a promising avenue for enhancing treatment outcomes in NSCLC. By combining the ancient wisdom of natural remedies with the precision of modern science, this study contributes to evolving cancer treatment paradigms, potentially mitigating the side effects associated with current therapies.
PubMed: 38919748
DOI: 10.3389/fmolb.2024.1413214 -
Biodesign Research 2024Living cells are exquisitely tuned to sense and respond to changes in their environment. Repurposing these systems to create engineered biosensors has seen growing...
Living cells are exquisitely tuned to sense and respond to changes in their environment. Repurposing these systems to create engineered biosensors has seen growing interest in the field of synthetic biology and provides a foundation for many innovative applications spanning environmental monitoring to improved biobased production. In this review, we present a detailed overview of currently available biosensors and the methods that have supported their development, scale-up, and deployment. We focus on genetic sensors in living cells whose outputs affect gene expression. We find that emerging high-throughput experimental assays and evolutionary approaches combined with advanced bioinformatics and machine learning are establishing pipelines to produce genetic sensors for virtually any small molecule, protein, or nucleic acid. However, more complex sensing tasks based on classifying compositions of many stimuli and the reliable deployment of these systems into real-world settings remain challenges. We suggest that recent advances in our ability to precisely modify nonmodel organisms and the integration of proven control engineering principles (e.g., feedback) into the broader design of genetic sensing systems will be necessary to overcome these hurdles and realize the immense potential of the field.
PubMed: 38919711
DOI: 10.34133/bdr.0037 -
International Journal of General... 2024To explore the predictive factors and predictive model construction for the progression of prostate cancer bone metastasis to castration resistance.
Development and Validation of a Clinic Machine Learning Classifier for the Prediction of Risk Stratifications of Prostate Cancer Bone Metastasis Progression to Castration Resistance.
OBJECTIVE
To explore the predictive factors and predictive model construction for the progression of prostate cancer bone metastasis to castration resistance.
METHODS
Clinical data of 286 patients diagnosed with prostate cancer with bone metastasis, initially treated with endocrine therapy, and progressing to metastatic castration resistant prostate cancer (mCRPC) were collected. By comparing the differences in various factors between different groups with fast and slow occurrence of castration-resistant prostate cancer (CRPC). Kaplan-Meier survival analysis and COX multivariate risk proportional regression model were used to compare the differences in the time to progression to CRPC in different groups. The COX multivariate risk proportional regression model was used to evaluate the impact of candidate factors on the time to progression to CRPC and establish a predictive model. The accuracy of the model was then tested using receiver operating characteristic (ROC) curves and decision curve analysis (DCA).
RESULTS
The median time for 286 mCRPC patients to progress to CRPC was 17 (9.5-28.0) months. Multivariate analysis showed that the lowest value of PSA (PSA nadir), the time when PSA dropped to its lowest value (timePSA), and the number of BM, and LDH were independent risk factors for rapid progression to CRPC. Based on the four independent risk factors mentioned above, a prediction model was established, with the optimal prediction model being a random forest with area under curve (AUC) of 0.946[95% CI: 0.901-0.991] and 0.927[95% CI: 0.864-0.990] in the training and validation cohort, respectively.
CONCLUSION
After endocrine therapy, the PSA nadir, timePSA, the number of BM, and LDH are the main risk factors for rapid progression to mCRPC in patients with prostate cancer bone metastases. Establishing a CRPC prediction model is helpful for early clinical intervention decision-making.
PubMed: 38919704
DOI: 10.2147/IJGM.S465031 -
Journal of Pediatric Intensive Care Jun 2024This study aimed to create a pediatric sedation scoring system independent of the American Society of Anesthesiology Physical Status (ASA-PS) classification that is...
This study aimed to create a pediatric sedation scoring system independent of the American Society of Anesthesiology Physical Status (ASA-PS) classification that is predictive of adverse events, facilitates objective stratification, and resource allocation. Multivariable regression and machine learning algorithm analysis of 134,973 sedation encounters logged in to the Pediatric Sedation Research Consortium (PSRC) database between July 2007 and June 2011. Patient and procedure variables were correlated with adverse events with resultant -regression coefficients used to assign point values to each variable. Point values were then summed to create a risk assessment score. Validation of the model was performed with the 2011 to 2013 PSRC database followed by calculation of ROC curves and positive predictive values. Factors identified and resultant point values are as follows: 1 point: age ≤ 6 months, cardiac diagnosis, asthma, weight less than 5th percentile or greater than 95 , and computed tomography (CT) scan; 2 points: magnetic resonance cholangiopancreatography (MRCP) and weight greater than 99th percentile; 4 points: magnetic resonance imaging (MRI); 5 points: trisomy 21 and esophagogastroduodenoscopy (EGD); 7 points: cough at the time of examination; and 18 points: bronchoscopy. Sum of patient and procedural values produced total risk assessment scores. Total risk assessment score of 5 had a sensitivity of 82.69% and a specificity of 26.22%, while risk assessment score of 11 had a sensitivity of 12.70% but a specificity of 95.29%. Inclusion of ASA-PS value did not improve model sensitivity or specificity and was thus excluded. Higher risk assessment scores predicted increased likelihood of adverse events during sedation. The score can be used to triage patients independent of ASA-PS with site-specific cut-off values used to determine appropriate sedation resource allocation.
PubMed: 38919693
DOI: 10.1055/s-0042-1745831 -
Frontiers in Immunology 2024This study aims to develop and validate machine learning models to predict proliferative lupus nephritis (PLN) occurrence, offering a reliable diagnostic alternative...
OBJECTIVE
This study aims to develop and validate machine learning models to predict proliferative lupus nephritis (PLN) occurrence, offering a reliable diagnostic alternative when renal biopsy is not feasible or safe.
METHODS
This study retrospectively analyzed clinical and laboratory data from patients diagnosed with SLE and renal involvement who underwent renal biopsy at West China Hospital of Sichuan University between 2011 and 2021. We randomly assigned 70% of the patients to a training cohort and the remaining 30% to a test cohort. Various machine learning models were constructed on the training cohort, including generalized linear models (e.g., logistic regression, least absolute shrinkage and selection operator, ridge regression, and elastic net), support vector machines (linear and radial basis kernel functions), and decision tree models (e.g., classical decision tree, conditional inference tree, and random forest). Diagnostic performance was evaluated using ROC curves, calibration curves, and DCA for both cohorts. Furthermore, different machine learning models were compared to identify key and shared features, aiming to screen for potential PLN diagnostic markers.
RESULTS
Involving 1312 LN patients, with 780 PLN/NPLN cases analyzed. They were randomly divided into a training group (547 cases) and a testing group (233 cases). we developed nine machine learning models in the training group. Seven models demonstrated excellent discriminatory abilities in the testing cohort, random forest model showed the highest discriminatory ability (AUC: 0.880, 95% confidence interval(CI): 0.835-0.926). Logistic regression had the best calibration, while random forest exhibited the greatest clinical net benefit. By comparing features across various models, we confirmed the efficacy of traditional indicators like anti-dsDNA antibodies, complement levels, serum creatinine, and urinary red and white blood cells in predicting and distinguishing PLN. Additionally, we uncovered the potential value of previously controversial or underutilized indicators such as serum chloride, neutrophil percentage, serum cystatin C, hematocrit, urinary pH, blood routine red blood cells, and immunoglobulin M in predicting PLN.
CONCLUSION
This study provides a comprehensive perspective on incorporating a broader range of biomarkers for diagnosing and predicting PLN. Additionally, it offers an ideal non-invasive diagnostic tool for SLE patients unable to undergo renal biopsy.
Topics: Humans; Lupus Nephritis; Female; Male; Machine Learning; Adult; Retrospective Studies; Middle Aged; Biomarkers; Young Adult
PubMed: 38919623
DOI: 10.3389/fimmu.2024.1413569