Use of CHAID decision trees to formulate pathways for the early detection of metabolic syndrome in young adults. Comput Math Methods Med. 2014;2014:242717. [PMID: 24817904 PMCID: PMC4003739 DOI: 10.1155/2014/242717]

Evidence from the large VALIGA cohort validates the subclassification of focal segmental glomerulosclerosis in IgA nephropathy

Evidence from the Oxford IgA nephropathy (IgAN) cohort supports the clinical value of subclassifying focal segmental glomerulosclerosis lesions (S1). Using the larger Validation in IgA (VALIGA) study cohort, we investigated the association between podocytopathic changes and higher proteinuria, kidney outcome and response to immunosuppressive therapy. All biopsies were evaluated for glomeruli with segmental capillary occlusion by matrix ("not otherwise specified", NOS lesion), simple capsular adhesion without capillary occlusion (Adh), tip lesions, and podocyte hypertrophy (PH). S1 required a NOS lesion and/or Adh. A Chi-Squared Automatic Interaction Detection method was used to identify subgroups of FSGS lesions associated with distinctive proteinuria at biopsy. We assessed survival from a combined event (kidney failure or 50% decline in estimated glomerular filtration rate). Finally, we evaluated within each subgroup if immunosuppression was associated with a favorable outcome using propensity analysis. In 1147 patients, S1 was found in 70% of biopsies. Subclassification found NOS lesions in 44%, Adh in 59%, PH in 13%, and tip lesions in 3%, with much overlap. Four subgroups were identified with progressively higher proteinuria: from lowest, S1 without NOS, S1 with NOS but without Adh/PH, to highest, S1 with NOS and Adh but without PH, and S1 with NOS and PH. These four subgroups showed progressively worse kidney survival. Immunosuppression was associated with a better outcome only in the two highest proteinuria subgroups. Propensity analysis in these two groups, adjusted for clinical and pathological findings, found a significantly reduced time-dependent hazard of combined outcome with corticosteroids. Podocyte hypertrophy and glomeruli with simple adhesions appeared to reflect active lesions associated with a response to corticosteroids, while other S1 lesions defined chronicity. Thus, our findings support subclassifying S1 lesions in IgAN.

Employing broad learning and non-invasive risk factor to improve the early diagnosis of metabolic syndrome

Metabolic syndrome (MetS) as a multifactorial disease is highly prevalent in countries and individuals. Monitoring the conventional risk factors (CRFs) would be a cost-effective strategy to target the increasing prevalence of MetS and the potential of noninvasive CRF for precisely detection of MetS in the early stage remains to be explored. From large-scale multicenter MetS clinical dataset, we discover 15 non-invasive CRFs which have strong relevance with MetS and first propose a broad learning-based approach named Genetic Programming Collaborative-competitive Broad Learning System (GP-CCBLS) with noninvasive CRF for early detection of MetS. The proposed GP-CCBLS model can significantly boost the detection performance and achieve the accuracy of 80.54%. This study supports the potential clinical validity of noninvasive CRF to complement general diagnostic criteria for early detecting the MetS and also illustrates possible strength of broad learning in disease diagnosis comparing with other machine learning approaches.

Linear Regression Modeling Based Scoring System to Reduce Benign Breast Biopsies Using Multi-parametric US with Color Doppler and SWE

RATIONALE AND OBJECTIVES

To develop a simple ultrasound (US) based scoring system to reduce benign breast biopsies.

MATERIALS AND METHODS

Women with BI-RADS 4 or 5 breast lesions underwent shear-wave elastography (SWE) imaging before biopsy. Standard US and color Doppler US (CDUS) parameters were recorded, and the size ratio (SzR=longest/shortest diameter) was calculated. Measured/calculated SWE parameters were minimum (SWVMin) and maximum (SWVMax) shear velocity, velocity heterogeneity (SWVH=SWVMax-SWVMin), velocity ratio (SWVR=SWVMin/SWVMax), and normalized SWVR (SWVRn=(SWVMax-SWVMin)/SWVMin). Linear regression analysis was performed by converting continuous parameters into categorical corresponding equivalents using decision tree analyses. Linear regression models were fitted using stepwise regression analysis and optimal coefficients for the predictors in the models were determined. A scoring model was devised from the results and validated using a different data set from another center consisting of 187 cases with BI-RADS 3, 4, and 5 lesions.

RESULTS

A total of 418 lesions (238 benign, 180 malignant) were analyzed. US and CDUS parameters exhibited poor (AUC=0.592-0.696), SWE parameters exhibited poor-good (AUC=0.607-0.816) diagnostic performance in benign/malignant discrimination. Linear regression models of US+CDUS and US+SWE parameters revealed an AUC of 0.819 and 0.882, respectively. The developed scoring system could have avoided biopsy in 37.8% of benign lesions while missing 1.1% of malignant lesions. The scoring system was validated with a 100% NPV rate with a specificity of 74.6%.

CONCLUSION

The linear regression model using US+SWE parameters performed better than any single parameter alone. The developed scoring method could lead to a significant decrease in benign biopsies.

Evaluating machine learning-powered classification algorithms which utilize variants in the GCKR gene to predict metabolic syndrome: Tehran Cardio-metabolic Genetics Study

BACKGROUND

Metabolic syndrome (MetS) is a prevalent multifactorial disorder that can increase the risk of developing diabetes, cardiovascular diseases, and cancer. We aimed to compare different machine learning classification methods in predicting metabolic syndrome status as well as identifying influential genetic or environmental risk factors.

METHODS

This candidate gene study was conducted on 4756 eligible participants from the Tehran Cardio-metabolic Genetic study (TCGS). We compared predictive models using logistic regression (LR), Random Forest (RF), decision tree (DT), support vector machines (SVM), and discriminant analyses. Demographic and clinical features, as well as variables regarding common GCKR gene polymorphisms, were included in the models. We used a 10-repeated tenfold cross-validation to evaluate model performance.

RESULTS

50.6% of participants had MetS. MetS was significantly associated with age, gender, schooling years, BMI, physical activity, rs780094, and rs780093 (P < 0.05) as indicated by LR. RF showed the best performance overall (AUC-ROC = 0.804, AUC-PR = 0.776, and Accuracy = 0.743) and indicated BMI, physical activity, and age to be the most influential model features. According to the DT, a person with BMI < 24 and physical activity < 8.8 possesses a 4% chance for MetS. In contrast, a person with BMI ≥ 25, physical activity < 2.7, and age ≥ 33, has 77% probability of suffering from MetS.

CONCLUSION

Our findings indicated that, on average, machine learning models outperformed conventional statistical approaches for patient classification. These well-performing models may be used to develop future support systems that use a variety of data sources to identify persons at high risk of getting MetS.

Risk factors, time to onset and recurrence of delirium in a mixed medical-surgical ICU population: A secondary analysis using Cox and CHAID decision tree modeling

A retrospective secondary analysis of 4,200 patients was collected from two academic medical centers. Delirium was assessed using the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) in all patients. Univariate and multivariate Cox models, logistic regression analysis, and Chi-square Automatic Interaction Detector (CHAID) decision tree modeling were used to explore delirium risk factors. Increased delirium risk was associated with exposed only to artificial light (AL) hazard ratio (HR) 1.84 (95 % CI: 1.66-2.044, P<0.001), physical restraint application 1.11 (95 % CI: 1.001-1.226, P=0.049), and high nursing care requirements (>8 hours per 8-hour shift) 1.18 (95 % CI: 1.048-1.338, P=0.007). Delirium incidence was inversely associated with greater family engagement 0.092 (95 % CI: 0.014-0.596, P=0.012), low staff burnout and anticipated turnover scores 0.093 (95 % CI: 0.014-0.600, P=0.013), non-ICU length-of-stay (LOS)<15 days 0.725 (95 % CI: 0.655-0.804, P<0.001), and ICU LOS ≤15 days 0.509 (95 % CI: 0.456-0.567, P<0.001). CHAID modeling indicated that AL exposure and age <65 years were associated with a high risk of delirium incidence, whereas SOFA score ≤11, APACHE IV score >15 and natural light (NL) exposure were associated with moderate risk, and female sex was associated with low risk. More rapid time to delirium onset correlated with baseline sleep disturbance (P=0.049), high nursing care requirements (P=0.019), and prolonged ICU and non-ICU hospital LOS (P<0.001). Delirium recurrence correlated with age >65 years (HR 2.198; 95 % CI: 1.101-4.388, P=0.026) and high nursing care requirements (HR 1.978, 95 % CI: 1.096-3.569), with CHAID modeling identifying AL exposure (P<0.001) and age >65 years (P=0.032) as predictive variables. Development of ICU delirium correlated with application of physical restraints, high nursing care requirements, prolonged ICU and non-ICU LOS, exposure exclusively to AL (rather than natural), less family engagement, and greater staff burnout and anticipated turnover scores. ICU delirium occurred more rapidly in patients with baseline sleep disturbance, and recurrence correlated with the presence of delirium on ICU admission, exclusive AL exposure, and high nursing care requirements.

Prediction of neonatal deaths in NICUs: development and validation of machine learning models

Background

Prediction of neonatal deaths in NICUs is important for benchmarking and evaluating healthcare services in NICUs. Application of machine learning techniques can improve physicians’ ability to predict the neonatal deaths. The aim of this study was to present a neonatal death risk prediction model using machine learning techniques.

Methods

This study was conducted in Tehran, Iran in two phases. Initially, important risk factors in neonatal death were identified and then several machine learning models including Artificial Neural Network (ANN), decision tree (Random Forest (RF), C5.0 and CHART tree), Support Vector Machine (SVM), Bayesian Network and Ensemble models were developed. Finally, we prospectively applied these models to predict neonatal death in a NICU and followed up the neonates to compare the outcomes of these neonates with real outcomes.

Results

17 factors were considered important in neonatal mortality prediction. The highest Area Under the Curve (AUC) was achieved for the SVM and Ensemble models with 0.98. The best precision and specificity were 0.98 and 0.94, respectively for the RF model. The highest accuracy, sensitivity and F-score were achieved for the SVM model with 0.94, 0.95 and 0.96, respectively. The best performance of models in prospective evaluation was for the ANN, C5.0 and CHAID tree models.

Conclusion

Using the developed machine learning models can help physicians predict the neonatal deaths in NICUs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12911-021-01497-8.

New Simplified Diagnostic Decision Trees for the Detention of Metabolic Syndrome in the Elderly

BACKGROUND

A new simplified method for the detention of metabolic syndrome (MetS) is proposed using two variables (anthropometric and minimally invasive).

METHODS

A study of MetS prevalence was made on a sample of 361 older people. The anthropometric variables analyzed were: blood pressure, body mass index, waist circumference (WC), waist-height ratio, body fat percentage, and waist-hip ratio. A crude and adjusted binary logistic regression was performed, and receiver operating characteristic curves were obtained for determining the predictive capacity of those variables. For the new detection method, decision trees were employed using automatic detection by interaction through Chi-square.

RESULTS

The prevalence of the MetS was of 43.7%. The final decision trees uses WC and basal glucose (BG), whose cutoff values were: for men, WC ≥ 102.5 cm and BG > 98 mg/dL (sensitivity = 67.1%, specificity = 90.3%, positive predictive value = 85%, validity index = 79.9%); and for women, WC ≥ 92.5 cm and BG ≥ 97 mg/dL (sensitivity = 65.9%, specificity = 92.7%, positive predictive value = 87.1%, validity index = 81.3%). In older women the best predictive value of MetS was a WC of 92.5 cm.

CONCLUSIONS

It is possible to make a simplified diagnosis of MetS in older people using the WC and basal capillary glucose, with a high diagnostic accuracy and whose use could be recommended in the resource-poor health areas. A new cutting point in older women for the WC should be valued.

Research on diagnosis-related group grouping of inpatient medical expenditure in colorectal cancer patients based on a decision tree model

BACKGROUND

In 2018, the diagnosis-related groups prospective payment system (DRGs-PPS) was introduced in a trial operation in Beijing according to the requirements of medical and health reform. The implementation of the system requires that more than 300 disease types pay through the DRGs-PPS for medical insurance. Colorectal cancer (CRC), as a common malignant tumor with high prevalence in recent years, was among the 300 disease types.

AIM

To investigate the composition and factors related to inpatient medical expenditure in CRC patients based on disease DRGs, and to provide a basis for the rational economic control of hospitalization expenses for the diagnosis and treatment of CRC.

METHODS

The basic material and cost data for 1026 CRC inpatients in a Grade-A tertiary hospital in Beijing during 2014-2018 were collected using the medical record system. A variance analysis of the composition of medical expenditure was carried out, and a multivariate linear regression model was used to select influencing factors with the greatest statistical significance. A decision tree model based on the exhaustive χ² automatic interaction detector (E-CHAID) algorithm for DRG grouping was built by setting chosen factors as separation nodes, and the payment standard of each diagnostic group and upper limit cost were calculated. The correctness and rationality of the data were re-evaluated and verified by clinical practice.

RESULTS

The average hospital stay of the 1026 CRC patients investigated was 18.5 d, and the average hospitalization cost was 57872.4 RMB yuan. Factors including age, gender, length of hospital stay, diagnosis and treatment, as well as clinical operations had significant influence on inpatient expenditure (P < 0.05). By adopting age, diagnosis, treatment, and surgery as the grouping nodes, a decision tree model based on the E-CHAID algorithm was established, and the CRC patients were divided into 12 DRG cost groups. Among these 12 groups, the number of patients aged ≤ 67 years, and underwent surgery and chemotherapy or radiotherapy was largest; while patients aged > 67 years, and underwent surgery and chemotherapy or radiotherapy had the highest medical cost. In addition, the standard cost and upper limit cost in the 12 groups were calculated and re-evaluated.

CONCLUSION

It is important to strengthen the control over the use of drugs and management of the hospitalization process, surgery, diagnosis and treatment to reduce the economic burden on patients. Tailored adjustments to medical payment standards should be made according to the characteristics and treatment of disease types to improve the comprehensiveness and practicability of the DRGs-PPS.

Educational intervention improves fruit and vegetable intake in young adults with metabolic syndrome components

The FRUVEDomics study investigates the effect of a diet intervention focused on increasing fruit and vegetable intake on the gut microbiome, and cardiovascular health of young adults with/at risk for Metabolic Syndrome (MetS). It was hypothesized the recommended diet would result in metabolic and gut microbiome changes. The 9-week dietary intervention adhered to the USDA Dietary Guidelines for Americans and focused on increasing fruit and vegetable intake to equal half of the diet. Seventeen eligible young adults with/or at high risk of MetS, consented and completed preintervention and postintervention measurements, including anthropometric, body composition, cardiovascular, complete blood lipid panel, and collection of stool sample for microbial analysis. Participants attended weekly consultations to assess food logs, food receipts, and adherence to the diet. Following intention-to-treat guidelines all 17 individuals were included in the dietary, clinical, and anthropometric analysis. Fruit and vegetable intake increased from 1.6 to 3.4 cups of fruits and vegetables (P < .001) daily. Total fiber (P = .02) and insoluble fiber (P < .0001) also increased. Clinical laboratory changes included an increase in sodium (P = .0006) and low-density lipoprotein cholesterol (P = .04). In the fecal microbiome, Erysipelotrichaceae (phylum Firmicutes) decreased (log2 fold change: −1.78, P = .01) and Caulobacteraceae (phylum Proteobacteria) increased (log2 fold change = 1.07, P = .01). Implementing a free living 9-week diet, with intensive education and accountability, gave young adults at high risk for/or diagnosed with MetS the knowledge, skills, and feedback to improve diet. To yield greater impact a longer diet intervention may be needed in this population.

Validation of a non-invasive method for the early detection of metabolic syndrome: a diagnostic accuracy test in a working population

OBJECTIVES

A non-invasive method for the early detection of metabolic syndrome (NIM-MetS) using only waist-to-height ratio (WHtR) and blood pressure (BP) has recently been published, with fixed cut-off values for gender and age. The aim of this study was to validate this method in a large sample of Spanish workers.

DESIGN

A diagnostic test accuracy to assess the validity of the method was performed.

SETTING

Occupational health services.

PARTICIPANTS

The studies were conducted in 2012-2016 on a sample of 60 799 workers from the Balearic Islands (Spain).

INTERVENTIONS

The NCEP-ATP III criteria were used as the gold standard. NIM-MetS has been devised using classification trees (the χ² automatic interaction detection method).

MAIN OUTCOME MEASURES

Anthropometric and biochemical variables to diagnose MetS. Sensitivity, specificity, validity index and Youden Index were determined to analyse the accuracy of the diagnostic test (NIM-MetS).

RESULTS

With regard to the validation of the method, sensitivity was 54.7%, specificity 94.9% and the Validity Index 91.2%. The cut-off value for WHtR was 0.54, ranging from 0.51 (lower age group) to 0.56 (higher age group). Variables more closely associated with MetS were WHtR (area under the curve (AUC)=0.85; 95% CI 0.84 to 0.86) and systolic BP (AUC=0.79; 95% CI 0.78 to 0.80)). The final cut-off values for the non-invasive method were WHtR ≥0.56 and BP ≥128/80 mm Hg, which includes four levels of MetS risk (very low, low, moderate and high).

CONCLUSIONS

The analysed method has shown a high validity index (higher than 91%) for the early detection of MetS. It is a non-invasive method that is easy to apply and interpret in any healthcare setting. This method provides a scale of MetS risk which allows more accurate detection and more effective intervention.

Factors related to limitation of life support within 48h of intensive care unit admission: A multicenter study

OBJECTIVE

To determine factors related to limitations on life support within 48h of intensive care unit (ICU) admission.

STUDY DESIGN

Prospective multicenter study.

SETTING

Eleven ICUs.

PATIENTS

All patients who died and/or had limitations on life support after ICU admission during a four-month period.

VARIABLES

Patient characteristics, hospital characteristics, characteristics of limitations on life support. Time-to-first-limitation was classified as early (<48h of admission) or late (≥48h). We performed univariate, multivariate analyses and CHAID (chi-square automatic interaction detection) analysis of variables associated with limitation of life support within 48h of ICU admission.

RESULTS

3335 patients were admitted; 326 (9.8%) had limitations on life support. A total of 344 patients died; 247 (71.8%) had limitations on life support (range among centers, 58.6%-84.2%). The median (p25-p75) time from admission to initial limitation was 2 (0-7) days. CHAID analysis found that the modified Rankin score was the variable most closely related with early limitations. Among patients with Rankin >2, early limitations were implemented in 71.7% (OR=2.5; 95% CI: 1.5-4.4) and lung disease was the variable most strongly associated with early limitations (OR=12.29; 95% CI: 1.63-255.91). Among patients with Rankin ≤2, 48.8% had early limitations; patients admitted after emergency surgery had the highest rate of early limitations (66.7%; OR=2.4; 95% CI: 1.1-5.5).

CONCLUSION

Limitations on life support are common, but the practice varies. Quality of life has the greatest impact on decisions to limit life support within 48h of admission.

Data mining for the identification of metabolic syndrome status

Metabolic syndrome (MS) is a condition associated with metabolic abnormalities that are characterized by central obesity (e.g. waist circumference or body mass index), hypertension (e.g. systolic or diastolic blood pressure), hyperglycemia (e.g. fasting plasma glucose) and dyslipidemia (e.g. triglyceride and high-density lipoprotein cholesterol). It is also associated with the development of diabetes mellitus (DM) type 2 and cardiovascular disease (CVD). Therefore, the rapid identification of MS is required to prevent the occurrence of such diseases. Herein, we review the utilization of data mining approaches for MS identification. Furthermore, the concept of quantitative population-health relationship (QPHR) is also presented, which can be defined as the elucidation/understanding of the relationship that exists between health parameters and health status. The QPHR modeling uses data mining techniques such as artificial neural network (ANN), support vector machine (SVM), principal component analysis (PCA), decision tree (DT), random forest (RF) and association analysis (AA) for modeling and construction of predictive models for MS characterization. The DT method has been found to outperform other data mining techniques in the identification of MS status. Moreover, the AA technique has proved useful in the discovery of in-depth as well as frequently occurring health parameters that can be used for revealing the rules of MS development. This review presents the potential benefits on the applications of data mining as a rapid identification tool for classifying MS.

Identification of Risk Factors of Multidrug-Resistant Tuberculosis by using Classification Tree Method

Multidrug-resistant tuberculosis (MDR-TB) has become a major public health problem. We tried to apply the classification tree model in building and evaluating a risk prediction model for MDR-TB. In this case-control study, 74 newly diagnosed MDR-TB patients served as the case group, and 95 patients without TB from the same medical institution served as the control group. The classification tree model was built using Chi-square Automatic Interaction Detectormethod and evaluated by income diagram, index map, risk statistic, and the area under receiver operating characteristic (ROC) curve. Four explanatory variables (history of exposure to TB patients, family with financial difficulties, history of other chronic respiratory diseases, and history of smoking) were included in the prediction model. The risk statistic of misclassification probability of the model was 0.160, and the area under ROC curve was 0.838 (P < 0.01). These suggest that the classification tree model works well for predicting MDR-TB. Classification tree model can not only predict the risk of MDR-TB effectively but also can reveal the interactions among variables.

A case study for a psychographic-behavioral segmentation approach for targeted demand generation in voluntary medical male circumcision

Public health programs are starting to recognize the need to move beyond a one-size-fits-all approach in demand generation, and instead tailor interventions to the heterogeneity underlying human decision making. Currently, however, there is a lack of methods to enable such targeting. We describe a novel hybrid behavioral-psychographic segmentation approach to segment stakeholders on potential barriers to a target behavior. We then apply the method in a case study of demand generation for voluntary medical male circumcision (VMMC) among 15-29 year-old males in Zambia and Zimbabwe. Canonical correlations and hierarchical clustering techniques were applied on representative samples of men in each country who were differentiated by their underlying reasons for their propensity to get circumcised. We characterized six distinct segments of men in Zimbabwe, and seven segments in Zambia, according to their needs, perceptions, attitudes and behaviors towards VMMC, thus highlighting distinct reasons for a failure to engage in the desired behavior.

Decision-Tree Analysis for Predicting First-Time Pass/Fail Rates for the NCLEX-RN® in Associate Degree Nursing Students

BACKGROUND

Little evidence shows the use of decision-tree algorithms in identifying predictors and analyzing their associations with pass rates for the NCLEX-RN(®) in associate degree nursing students. This longitudinal and retrospective cohort study investigated whether a decision-tree algorithm could be used to develop an accurate prediction model for the students' passing or failing the NCLEX-RN.

METHOD

This study used archived data from 453 associate degree nursing students in a selected program. The chi-squared automatic interaction detection analysis of the decision trees module was used to examine the effect of the collected predictors on passing/failing the NCLEX-RN.

RESULTS

The actual percentage scores of Assessment Technologies Institute®'s RN Comprehensive Predictor(®) accurately identified students at risk of failing. The classification model correctly classified 92.7% of the students for passing.

CONCLUSION

This study applied the decision-tree model to analyze a sequence database for developing a prediction model for early remediation in preparation for the NCLEXRN. [J Nurs Educ. 2016;55(8):454-457.].