Application of radiomics model based on lumbar computed tomography in diagnosis of elderly osteoporosis

A metabolic bone disease characterized by decreased bone formation and increased bone resorption is osteoporosis. It can cause pain and fracture of patients. The elderly are prone to osteoporosis and are more vulnerable to osteoporosis. In this study, radiomics are extracted from computed tomography (CT) images to screen osteoporosis in the elderly. Collect the plain scan CT images of lumbar spine, cut the region of interest of the image and extract radiomics features, use Lasso regression to screen variables and adjust complexity, use python language to model random forests, support vector machines, K nearest neighbor, and finally use receiver operating characteristic curve to evaluate the performance of the model, including precision, recall, accuracy and area under the curve (AUC). For the model, 14 radiolomics features were selected. The diagnosis performance of random forest model and support vector machine is good, all around 0.9. The AUC of K nearest neighbor model in training set and test set is 0.828 and 0.796, respectively. We selected the plain scan CT images of the elderly lumbar spine to build radiomics features model, which has good diagnostic performance and can be used as a tool to assist the diagnosis of osteoporosis in the elderly.

A systematic evaluation and meta-analysis of early prediction of post-thrombotic syndrome

Objective

Post-thrombotic syndrome (PTS) is the most common long-term complication in patients with deep venous thrombosis, and the prevention of PTS remains a major challenge in clinical practice. Some studies have explored early predictors and constructed corresponding prediction models, whereas their specific application and predictive value are controversial. Therefore, we conducted this systematic evaluation and meta-analysis to investigate the incidence of PTS and the feasibility of early prediction.

Methods

We systematically searched databases of PubMed, Embase, Cochrane and Web of Science up to April 7, 2023. Newcastle-Ottawa Scale (NOS) was used to evaluate the quality of the included articles, and the OR values of the predictors in multi-factor logistic regression were pooled to assess whether they could be used as effective independent predictors.

Results

We systematically included 20 articles involving 8,512 subjects, with a predominant onset of PTS between 6 and 72 months, with a 2-year incidence of 37.5% (95% CI: 27.8-47.7%). The results for the early predictors were as follows: old age OR = 1.840 (95% CI: 1.410-2.402), obesity or overweight OR = 1.721 (95% CI: 1.245-2.378), proximal deep vein thrombosis OR = 2.335 (95% CI: 1.855-2.938), history of venous thromboembolism OR = 3.593 (95% CI: 1.738-7.240), history of smoking OR = 2.051 (95% CI: 1.305-3.224), varicose veins OR = 2.405 (95% CI: 1.344-4.304), and baseline Villalta score OR = 1.095(95% CI: 1.056-1.135). Meanwhile, gender, unprovoked DVT and insufficient anticoagulation were not independent predictors. Seven studies constructed risk prediction models. In the training set, the c-index of the prediction models was 0.77 (95% CI: 0.74-0.80) with a sensitivity of 0.75 (95% CI: 0.68-0.81) and specificity of 0.69 (95% CI: 0.60-0.77). In the validation set, the c-index, sensitivity and specificity of the prediction models were 0.74(95% CI: 0.69-0.79), 0.71(95% CI: 0.64-0.78) and 0.72(95% CI: 0.67-0.76), respectively.

Conclusions

With a high incidence after venous thrombosis, PTS is a complication that cannot be ignored in patients with venous thrombosis. Risk prediction scoring based on early model construction is a feasible option, which helps to identify the patient's condition and develop an individualized prevention program to reduce the risk of PTS.

Evaluation of EuroQol Valuation Technology (EQ-VT) Designs to Generate National Value Sets: Learnings from the Development of an EQ-5D Value Set for India Using an Extended Design (DEVINE) Study

INTRODUCTION

Countries develop their EQ-5D-5L value sets using the EuroQol Valuation Technology (EQ-VT) protocol. This study aims to assess if extension in the conventional EQ-VT design can lead to development of value sets with improved precision.

METHODS

A cross-sectional survey was undertaken in a representative sample of 3,548 adult respondents, selected from 5 different states of India using a multistage stratified random sampling technique. A novel extended EQ-VT design was created that included 18 blocks of 10 health states, comprising 150 unique health states and 135 observations per health state. In addition to the standard EQ-VT design, which is based on 86 health states and 100 observations per health state, 3 extended designs were assessed for their predictive performance. The extended designs were created by 1) increasing the number of observations per health state in the design, 2) increasing the number of health states in the design, and 3) implementing both 1) and 2) at the same time. Subsamples of the data set were created for separate designs. The root mean squared error (RMSE) and mean absolute error (MAE) were used to measure the predictive accuracy of the conventional and extended designs.

RESULTS

The average RMSE and MAE for the standard EQ-VT design were 0.055 and 0.041, respectively, for the 150 health states. All 3 types of design extensions showed lower RMSE and MAE values as compared with the standard design and hence yielded better predictive performance. RMSE and MAE were lowest (0.051 and 0.039, respectively) for the designs that use a greater number of health states. Extending the design with inclusion of more health states was shown to improve the predictive performance even when the sample size was fixed at 1,000.

CONCLUSION

Although the standard EQ-VT design performs well, its prediction accuracy can be further improved by extending its design. The addition of more health states in EQ-VT is more beneficial than increasing the number of observations per health state.

HIGHLIGHTS

The EQ-5D-5L value sets are developed using the standardized EuroQol Valuation Technology (EQ-VT) protocol. This is the first study to empirically assess how much can be gained from extending the standard EQ-VT design in terms of sample size and/or health states. It not only presents useful insights into the performance of the standard design of the EQ-VT but also tests the potential extensions in the standard EQ-VT design in terms of increasing the health states to be directly valued as well as the number of observations recorded to predict the utility value of each of these health states.The study demonstrates that the standard EQ-VT design performs good, and an extension in the design of the standard EQ-VT can lead to further improvement in its performance. The addition of more health states in EQ-VT is more beneficial than increasing the number of observations per health state. Extending the design with inclusion of more health states marginally improves the predictive performance even when the sample size was fixed at 1,000.The findings of the study will streamline the systematic process for generating precise EQ-5D-5L value sets, thus facilitating the conduct of credible, transparent, and robust outcome valuation in health technology assessments.

Research Progress and New Ideas on the Theory and Methodology of Water Quality Criteria for the Protection of Aquatic Organisms

Water quality criteria (WQC) for the protection of aquatic organisms mainly focus on the maximum threshold values of the pollutants that do not have harmful effects on aquatic organisms. The WQC value is the result obtained based on scientific experiments in the laboratory and data fitting extrapolation and is the limit of the threshold value of pollutants or other harmful factors in the water environment. Until now, many studies have been carried out on WQC for the protection of aquatic organisms internationally, and several countries have also issued their own relevant technical guidelines. Thus, the WQC method for the protection of aquatic organisms has been basically formed, with species sensitivity distribution (SSD) as the main method and the assessment factor (AF) as the auxiliary method. In addition, in terms of the case studies on WQC, many scholars have conducted relevant case studies on various pollutants. At the national level, several countries have also released WQC values for typical pollutants. This study systematically discusses the general situation, theoretical methodology and research progress of WQC for the protection of aquatic organisms, and deeply analyzes the key scientific issues that need to be considered in the research of WQC. Furthermore, combined with the specific characteristics of the emerging pollutants, some new ideas and directions for future WQC research for the protection of aquatic organisms are also proposed.

Machine learning prediction of motor function in chronic stroke patients: a systematic review and meta-analysis

Background

Recent studies have reported that machine learning (ML), with a relatively strong capacity for processing non-linear data and adaptive ability, could improve the accuracy and efficiency of prediction. The article summarizes the published studies on ML models that predict motor function 3-6 months post-stroke.

Methods

A systematic literature search was conducted in PubMed, Embase, Cochorane and Web of Science as of April 3, 2023 for studies on ML prediction of motor function in stroke patients. The quality of the literature was assessed using the Prediction model Risk Of Bias Assessment Tool (PROBAST). A random-effects model was preferred for meta-analysis using R4.2.0 because of the different variables and parameters.

Results

A total of 44 studies were included in this meta-analysis, involving 72,368 patients and 136 models. Models were categorized into subgroups according to the predicted outcome Modified Rankin Scale cut-off value and whether they were constructed based on radiomics. C-statistics, sensitivity, and specificity were calculated. The random-effects model showed that the C-statistics of all models were 0.81 (95% CI: 0.79; 0.83) in the training set and 0.82 (95% CI: 0.80; 0.85) in the validation set. According to different Modified Rankin Scale cut-off values, C-statistics of ML models predicting Modified Rankin Scale>2(used most widely) in stroke patients were 0.81 (95% CI: 0.78; 0.84) in the training set, and 0.84 (95% CI: 0.81; 0.87) in the validation set. C-statistics of radiomics-based ML models in the training set and validation set were 0.81 (95% CI: 0.78; 0.84) and 0.87 (95% CI: 0.83; 0.90), respectively.

Conclusion

ML can be used as an assessment tool for predicting the motor function in patients with 3-6 months of post-stroke. Additionally, the study found that ML models with radiomics as a predictive variable were also demonstrated to have good predictive capabilities. This systematic review provides valuable guidance for the future optimization of ML prediction systems that predict poor motor outcomes in stroke patients.

Systematic review registration

https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42022335260, identifier: CRD42022335260.

China Requires a Sustainable Transition of Vegetable Supply from Area-Dependent to Yield-Dependent and Decreased Vegetable Loss and Waste

China, the largest country in vegetable supply, faces a transition to sustainable vegetable production to counteract resource waste and environmental pollution. However, there are knowledge gaps on the main constraints and how to achieve sustainable vegetable supply. Herein, we integrated the vegetable production and supply data in China, compared its current status with five horticulture-developed countries US, the Netherlands, Greece, Japan and South Korea, using data from the Food and Agriculture Organization (FAO) and National Bureau of Statistics of China, and predicted the vegetable supply in 2030 and 2050 by a model prediction. The vegetable supply in China increased from 592 g capita-1 d-1 in 1995 to 1262 g capita-1 d-1 in 2018 and will keep constant in 2030 and 2050. Compared to the five countries, the greater vegetable supply is primarily achieved by higher harvested areas rather than higher yield. However, it is predicted that the harvested areas will decrease by 13.6% and 24.7% in 2030 and 2050. Instead, steady increases in vegetable yield by 11.8% and 28.3% are predicted for this period. The high vegetable supply and greater vegetable preference indicated by the high vegetable-to-meat production ratio cannot guarantee recommended vegetable intake, potentially due to the high rate of vegetable loss and waste. Under the scenarios of decreased vegetable loss and waste, the harvested area will decrease by 37.3-67.2% in 2030 and 2050. This study points out that the sustainable transition of Chinese vegetable supply can be realized by enhancing yield and limiting vegetable loss and waste instead of expanding the harvested area.

Disease-specific data processing: An intelligent digital platform for diabetes based on model prediction and data analysis utilizing big data technology

Background

Artificial intelligence technology has become a mainstream trend in the development of medical informatization. Because of the complex structure and a large amount of medical data generated in the current medical informatization process, big data technology to assist doctors in scientific research and analysis and obtain high-value information has become indispensable for medical and scientific research.

Methods

This study aims to discuss the architecture of diabetes intelligent digital platform by analyzing existing data mining methods and platform building experience in the medical field, using a large data platform building technology utilizing the Hadoop system, model prediction, and data processing analysis methods based on the principles of statistics and machine learning. We propose three major building mechanisms, namely the medical data integration and governance mechanism (DCM), data sharing and privacy protection mechanism (DPM), and medical application and medical research mechanism (MCM), to break down the barriers between traditional medical research and digital medical research. Additionally, we built an efficient and convenient intelligent diabetes model prediction and data analysis platform for clinical research.

Results

Research results from this platform are currently applied to medical research at Shanghai T Hospital. In terms of performance, the platform runs smoothly and is capable of handling massive amounts of medical data in real-time. In terms of functions, data acquisition, cleaning, and mining are all integrated into the system. Through a simple and intuitive interface operation, medical and scientific research data can be processed and analyzed conveniently and quickly.

Conclusions

The platform can serve as an auxiliary tool for medical personnel and promote the development of medical informatization and scientific research. Also, the platform may provide the opportunity to deliver evidence-based digital therapeutics and support digital healthcare services for future medicine.

Automated machine learning-based model for the prediction of delirium in patients after surgery for degenerative spinal disease

OBJECTIVE

This study used machine learning algorithms to identify critical variables and predict postoperative delirium (POD) in patients with degenerative spinal disease.

METHODS

We included 663 patients who underwent surgery for degenerative spinal disease and received general anesthesia. The LASSO method was used to screen essential features associated with POD. Clinical characteristics, preoperative laboratory parameters, and intraoperative variables were reviewed and were used to construct nine machine learning models including a training set and validation set (80% of participants), and were then evaluated in the rest of the study sample (20% of participants). The area under the receiver-operating characteristic curve (AUROC) and Brier scores were used to compare the prediction performances of different models. The eXtreme Gradient Boosting algorithms (XGBOOST) model was used to predict POD. The SHapley Additive exPlanations (SHAP) package was used to interpret the XGBOOST model. Data of 49 patients were prospectively collected for model validation.

RESULTS

The XGBOOST model outperformed the other classifier models in the training set (area under the curve [AUC]: 92.8%, 95% confidence interval [CI]: 90.7%-95.0%), validation set (AUC: 87.0%, 95% CI: 80.7%-93.3%). This model also achieved the lowest Brier Score. Twelve vital variables, including age, serum albumin, the admission-to-surgery time interval, C-reactive protein level, hypertension, intraoperative blood loss, intraoperative minimum blood pressure, cardiovascular-cerebrovascular disease, smoking, alcohol consumption, pulmonary disease, and admission-intraoperative maximum blood pressure difference, were selected. The XGBOOST model performed well in the prospective cohort (accuracy: 85.71%).

CONCLUSION

A machine learning model and a web predictor for delirium after surgery for the degenerative spinal disease were successfully developed to demonstrate the extent of POD risk during the perioperative period, which could guide appropriate preventive measures for high-risk patients.

Machine Learning-Assisted Identification and Quantification of Hydroxylated Metabolites of Polychlorinated Biphenyls in Animal Samples

Laboratory studies of the disposition and toxicity of hydroxylated polychlorinated biphenyl (OH-PCB) metabolites are challenging because authentic analytical standards for most unknown OH-PCBs are not available. To assist with the characterization of these OH-PCBs (as methylated derivatives), we developed machine learning-based models with multiple linear regression (MLR) or random forest regression (RFR) to predict the relative retention times (RRT) and MS/MS responses of methoxylated (MeO-)PCBs on a gas chromatograph-tandem mass spectrometry system. The final MLR model estimated the retention times of MeO-PCBs with a mean absolute error of 0.55 min (n = 121). The similarity coefficients cos θ between the predicted (by RFR model) and experimental MS/MS data of MeO-PCBs were >0.95 for 92% of observations (n = 96). The levels of MeO-PCBs quantified with the predicted MS/MS response factors approximated the experimental values within a 2-fold difference for 85% of observations and 3-fold differences for all observations (n = 89). Subsequently, these model predictions were used to assist with the identification of OH-PCB 95 or OH-PCB 28 metabolites in mouse feces or liver by suggesting candidate ranking information for identifying the metabolite isomers. Thus, predicted retention and MS/MS response data can assist in identifying unknown OH-PCBs.

Bioaccumulation of Pyraoxystrobin and Its Predictive Evaluation in Zebrafish

This paper aims to understand the bioaccumulation of pyraoxystrobin in fish. Using a flow-through bioconcentration method, the bioconcentration factor (BCF) and clearance rate of pyraoxystrobin in zebrafish were measured. The measured BCF values were then compared to those estimated from three commonly used predication models. At the exposure concentrations of 0.1 μg/L and 1.0 μg/L, the maximum BCF values for pyraoxystrobin in fish were 820.8 and 265.9, and the absorption rate constants (K₁) were 391.0 d⁻¹ and 153.2 d⁻¹, respectively. The maximum enrichment occurred at 12 d of exposure. At the two test concentrations, the clearance rate constant (K₂) in zebrafish was 0.5795 and 0.4721, and the half-life (t_1/2) was 3.84 d and 3.33 d, respectively. The measured BCF values were close to those estimated from bioconcentration predication models.

Automated machine learning-based model predicts postoperative delirium using readily extractable perioperative collected electronic data

Objective

Postoperative delirium (POD) is a common postoperative complication that is relevant to poor outcomes. Therefore, it is critical to find effective methods to identify patients with high risk of POD rapidly. Creating a fully automated score based on an automated machine‐learning algorithm may be a method to predict the incidence of POD quickly.

Materials and methods

This is the secondary analysis of an observational study, including 531 surgical patients who underwent general anesthesia. The least absolute shrinkage and selection operator (LASSO) was used to screen essential features associated with POD. Finally, eight features (age, intraoperative blood loss, anesthesia duration, extubation time, intensive care unit [ICU] admission, mini‐mental state examination score [MMSE], Charlson comorbidity index [CCI], postoperative neutrophil‐to‐lymphocyte ratio [NLR]) were used to established models. Four models, logistic regression, random forest, extreme gradient boosted trees, and support vector machines, were built in a training set (70% of participants) and evaluated in the remaining testing sample (30% of participants). Multivariate logistic regression analysis was used to explore independent risk factors for POD further.

Results

Model 1 (logistic regression model) was found to outperform other classifier models in testing data (area under the curve [AUC] of 80.44%, 95% confidence interval [CI] 72.24%–88.64%) and achieve the lowest Brier Score as well. These variables including age (OR = 1.054, 95%CI: 1.017~1.093), extubation time (OR = 1.027, 95%CI: 1.012~1.044), ICU admission (OR = 2.238, 95%CI: 1.313~3.793), MMSE (OR = 0.929, 95%CI: 0.876~0.984), CCI (OR = 1.197, 95%CI: 1.038~1.384), and postoperative NLR (OR = 1.029, 95%CI: 1.002~1.057) were independent risk factors for POD in this study.

Conclusions

We have built and validated a high‐performing algorithm to demonstrate the extent to which patient risk changes of POD during the perioperative period, thus leading to a rational therapeutic choice.

Development and internal validation of a predictive model of cognitive decline 36 months following elective surgery

INTRODUCTION

Our goal was to determine if features of surgical patients, easily obtained from the medical chart or brief interview, could be used to predict those likely to experience more rapid cognitive decline following surgery.

METHODS

We analyzed data from an observational study of 560 older adults (≥70 years) without dementia undergoing major elective non-cardiac surgery. Cognitive decline was measured using change in a global composite over 2 to 36 months following surgery. Predictive features were identified as variables readily obtained from chart review or a brief patient assessment. We developed predictive models for cognitive decline (slope) and predicting dichotomized cognitive decline at a clinically determined cut.

RESULTS

In a hold-out testing set, the regularized regression predictive model achieved a root mean squared error (RMSE) of 0.146 and a model r-square (R2 ) of .31. Prediction of "rapid" decliners as a group achieved an area under the curve (AUC) of .75.

CONCLUSION

Some of our models could predict persons with increased risk for accelerated cognitive decline with greater accuracy than relying upon chance, and this result might be useful for stratification of surgical patients for inclusion in future clinical trials.

Factors Affecting the Number of Pollen Grains per Male Strobilus in Japanese Cedar (Cryptomeria japonica)

Japanese cedar (Cryptomeria japonica) is the most important timber species in Japan; however, its pollen is the primary cause of pollinosis in Japan. The total number of pollen grains produced by a single tree is determined by the number of male strobili (male flowers) and the number of pollen grains per male strobilus. While the number of male strobili is a visible and well-investigated trait, little is known about the number of pollen grains per male strobilus. We hypothesized that genetic and environmental factors affect the pollen number per male strobilus and explored the factors that affect pollen production and genetic variation among clones. We counted pollen numbers of 523 male strobili from 26 clones using a cell counter method that we recently developed. Piecewise Structural Equation Modeling (pSEM) revealed that the pollen number is mostly affected by genetic variation, male strobilus weight, and pollen size. Although we collected samples from locations with different environmental conditions, statistical modeling succeeded in predicting pollen numbers for different clones sampled from branches facing different directions. Comparison of predicted pollen numbers revealed that they varied >3-fold among the 26 clones. The determination of the factors affecting pollen number and a precise evaluation of genetic variation will contribute to breeding strategies to counter pollinosis. Furthermore, the combination of our efficient counting method and statistical modeling will provide a powerful tool not only for Japanese cedar but also for other plant species.

Construction of a prediction model and a prevention control system for cesarean section rate based on the Robson classification system

OBJECTIVE

To systematically explore the risk factors that influence cesarean section rate, and establish a prediction model to investigate a system effectively reducing cesarean section rates.

METHODS

This retrospective study was carried out in the medical institutions in Xingtai city, where cesarean section could be conducted. The data of parturients who gave birth to children in the past five years were collected using the hospital information system. Based on the Robson's ten group classification system, parturients were grouped. The difference of cesarean section rate in each group and its main influencing factors were then analyzed. The above factors and factors such as age, education background, and knowledge on childbirth were independent variables, while cesarean section was the dependent variable. A logistic regression model was constructed to determine the correlation between relevant influencing factors and cesarean section.

RESULTS

In the past 5 years, cesarean section rate in Xingtai city had been maintained at a relatively high level. Cesarean section rates in the R2 and R5 groups were the highest. Parity, fetal position, number of fetuses, and gestational weeks were all factors affecting cesarean section rate (all P < 0.01). After screening the above factors using logistic regression analysis, a regression equation was established: logistic (p) = -1.061 + 1.107 * parity + 0.196 * fetal position + 2.245 * number of fetuses - 0.070 * gestational week + 0.234 * age - 0.278 * education background + 0.623 * knowledge on childbirth.

CONCLUSION

The Robson classification system plays an important role in evaluating and supervising parturients' conditions. Based on the Robson classification system, we find that parity, fetal position, number of fetuses, and gestational weeks are the main factors influencing cesarean section rate. Using logistic regression analysis, a prediction model, with guiding significance on the control of cesarean section rate, is established.

Occurrence, Removal, and Mass Balance of Polycyclic Aromatic Hydrocarbons and Their Derivatives in Wastewater Treatment Plants in Northeast China

Polycyclic aromatic hydrocarbons (PAHs), 33 methylated PAHs (Me-PAHs), and 14 nitrated PAHs (NPAHs) were measured in wastewater treatment plants (WWTPs) to study the removal efficiency of these compounds through the WWTPs, as well as their source appointment and potential risk in the effluent. The concentrations of ∑PAHs, ∑Me-PAHs, and ∑NPAHs were 2.01–8.91, 23.0–102, and 6.21–171 µg/L in the influent, and 0.17–1.37, 0.06–0.41 and 0.01–2.41 µg/L in the effluent, respectively. Simple Treat 4.0 and meta-regression methods were applied to calculate the removal efficiencies (REs) for the 63 PAHs and their derivatives in 10 WWTPs and the results were compared with the monitoring data. Overall, the ranges of REs were 55.3–95.4% predicated by the Simple Treat and 47.5–97.7% by the meta-regression. The results by diagnostic ratios and principal component analysis PCA showed that “mixed source” biomass, coal composition, and petroleum could be recognized to either petrogenic or pyrogenic sources. The risk assessment of the effluent was also evaluated, indicating that seven carcinogenic PAHs, Benzo[a]pyrene, Dibenz[a,h]anthracene, and Benzo(a)anthracene were major contributors to the toxics equivalency concentrations (TEQs) in the effluent of WWTPs, to which attention should be paid.

Comparing the Predictive Effects of Patient Medication Adherence Indices in Electronic Health Record and Claims-Based Risk Stratification Models

Multiple indices are available to measure medication adherence behaviors. Medication adherence measures, however, have rarely been extracted from electronic health records (EHRs) for population-level risk predictions. This study assessed the value of medication adherence indices in improving predictive models of cost and hospitalization. This study included a 2-year retrospective cohort of patients younger than age 65 years with linked EHR and insurance claims data. Three medication adherence measures were calculated: medication regimen complexity index (MRCI), medication possession ratio (MPR), and prescription fill rate (PFR). The authors examined the effects of adding these measures to 3 predictive models of utilization: a demographics model, a conventional model (Charlson index), and an advanced diagnosis-based model. Models were trained using EHR and claims data. The study population had an overall MRCI, MPR, and PFR of 14.6 ± 17.8, .624 ± .310, and .810 ± .270, respectively. Adding MRCI and MPR to the demographic and the morbidity models using claims data improved forecasting of next-year hospitalization substantially (eg, AUC of the demographic model increased from .605 to .656 using MRCI). Nonetheless, such boosting effects were attenuated for the advanced diagnosis-based models. Although EHR models performed inferior to claims models, adding adherence indices improved EHR model performances at a larger scale (eg, adding MRCI increased AUC by 4.4% for the Charlson model using EHR data compared to 3.8% using claims). This study shows that medication adherence measures can modestly improve EHR- and claims-derived predictive models of cost and hospitalization in non-elderly patients; however, the improvements are minimal for advanced diagnosis-based models.

Machine Learning to Develop and Internally Validate a Predictive Model for Post-operative Delirium in a Prospective, Observational Clinical Cohort Study of Older Surgical Patients

BACKGROUND

Our objective was to assess the performance of machine learning methods to predict post-operative delirium using a prospective clinical cohort.

METHODS

We analyzed data from an observational cohort study of 560 older adults (≥ 70 years) without dementia undergoing major elective non-cardiac surgery. Post-operative delirium was determined by the Confusion Assessment Method supplemented by a medical chart review (N = 134, 24%). Five machine learning algorithms and a standard stepwise logistic regression model were developed in a training sample (80% of participants) and evaluated in the remaining hold-out testing sample. We evaluated three overlapping feature sets, restricted to variables that are readily available or minimally burdensome to collect in clinical settings, including interview and medical record data. A large feature set included 71 potential predictors. A smaller set of 18 features was selected by an expert panel using a consensus process, and this smaller feature set was considered with and without a measure of pre-operative mental status.

RESULTS

The area under the receiver operating characteristic curve (AUC) was higher in the large feature set conditions (range of AUC, 0.62-0.71 across algorithms) versus the selected feature set conditions (AUC range, 0.53-0.57). The restricted feature set with mental status had intermediate AUC values (range, 0.53-0.68). In the full feature set condition, algorithms such as gradient boosting, cross-validated logistic regression, and neural network (AUC = 0.71, 95% CI 0.58-0.83) were comparable with a model developed using traditional stepwise logistic regression (AUC = 0.69, 95% CI 0.57-0.82). Calibration for all models and feature sets was poor.

CONCLUSIONS

We developed machine learning prediction models for post-operative delirium that performed better than chance and are comparable with traditional stepwise logistic regression. Delirium proved to be a phenotype that was difficult to predict with appreciable accuracy.

Modeling and Prediction of Wearable Energy Harvesting Sliding Shoes for Metabolic Cost and Energy Rate Outside of the Lab

The recent explosion of wearable electronics has led to widespread interest in harvesting human movement energy, particularly during walking, for clinical and health applications. However, the amount of energy available to harvest and the required metabolic rate for wearable energy harvesting varies across subjects. In this paper, we utilize custom energy harvesting sliding shoes to develop and evaluate multivariate linear regression models to predict metabolic rate and energy harvesting rate during overground walking outside of the lab. Subjects performed 200 m self-selected normal and fast walking trials on flat ground with custom sliding shoes. Metabolic rate was measured with a portable breathing analysis system and energy harvesting rate was measured directly from the generator on the custom sliding shoes. Model performance was determined by comparing the difference between actual and predicted metabolic and energy harvesting rates. Overall, predictive modeling closely matched the actual values, and there was no statistical difference between actual and predicted average metabolic rate or between actual and predicted average energy harvesting rate. Energy harvesting sliding shoes could potentially be used for a variety of wearable devices to reduce onboard energy storage, and these findings could serve to inform expected energy harvesting rates and associated required metabolic cost for a diverse array of medical and health applications.

Globally Normal Bistable Motion Perception of Anisometropic Amblyopes May Profit From an Unusual Coding Mechanism

Anisometropic amblyopia is a neurodevelopmental disorder of the visual system. There is evidence that the neural deficits spread across visual areas, from the primary cortex up to higher brain areas, including motion coding structures such as MT. Here, we used bistable plaid motion to investigate changes in the underlying mechanisms of motion integration and segmentation and, thus, help us to unravel in more detail deficits in the amblyopic visual motion system. Our results showed that (1) amblyopes globally exhibited normal bistable perception in all viewing conditions compared to the control group and (2) decreased contrast led to a stronger increase in percept switches and decreased percept durations in the control group, while the amblyopic group exhibited no such changes. There were few differences in outcomes dependent upon the use of the weak eye, the strong eye, or both eyes for viewing the stimuli, but this was a general effect present across all subjects, not specific to the amblyopic group. To understand the role of noise and adaptation in such cases of bistable perception, we analyzed predictions from a model and found that contrast does indeed affect percept switches and durations as observed in the control group, in line with the hypothesis that lower stimulus contrast enhances internal noise effects. The combination of experimental and computational results presented here suggests a different motion coding mechanism in the amblyopic visual system, with relatively little effect of stimulus contrast on amblyopes' bistable motion perception.

Effect of Over-Tree Evaporative Cooling in Orchards on Microclimate and Accuracy of Insect Model Predictions

Orchard design and management practices can alter microclimate and, thus, potentially affect insect development. If sufficiently large, these deviations in microclimate can compromise the accuracy of phenology model predictions used in integrated pest management (IPM) programs. Sunburn causes considerable damage in the Pacific Northwest, United States, apple-producing region. Common prevention strategies include the use of fruit surface protectants, evaporative cooling (EC), or both. This study focused on the effect of EC on ambient temperatures and model predictions for four insects (codling moth, Cydia pomonella L.; Lacanobia fruitworm, Lacanobia subjuncta Grote and Robinson; oblique-banded leafroller, Choristoneura rosaceana Harris; and Pandemis leafroller, Pandemis pyrusana Kearfott). Over-tree EC was applied in July and August when daily maximum temperatures were predicted to be ≥30°C between 1200-1700 hours (15/15 min on/off interval) in 2011 and between 1200-1800 hours (15/10 min on/off interval, or continuous on) in 2012. Control plots were sprayed once with kaolin clay in early July. During interval and continuous cooling, over-tree cooling reduced average afternoon temperatures compared with the kaolin treatment by 2.1-3.2°C. Compared with kaolin-treated controls, codling moth and Lacanobia fruitworm egg hatch in EC plots was predicted to occur up to 2 d and 1 d late, respectively. The presence of fourth-instar oblique-banded leafroller and Pandemis leafroller was predicted to occur up to 2 d and 1 d earlier in EC plots, respectively. These differences in model predictions were negligible, suggesting that no adjustments in pest management timing are needed when using EC in high-density apple orchards.

Modelling HIV immune response and validation with clinical data

A system of ordinary differential equations is formulated to describe the pathogenesis of HIV infection, wherein certain features that have been shown to be important by recent experimental research are incorporated in the model. These include the role of CD4+ memory cells that serve as a major reservoir of latently infected cells, a critical role for T-helper cells in the generation of CD8 memory cells capable of efficient recall response, and stimulation by antigens other than HIV. A stability analysis illustrates the capability of this model in admitting multiple locally asymptotically stable (locally a.s.) off-treatment equilibria.We show that this more biologically detailed model can exhibit the phenomenon of transient viremia experienced by some patients on therapy with viral load levels suppressed below the detection limit. We also show that the loss of CD4+ T-cell help in the generation of CD8+ memory cells leads to larger peak values for the viral load during transient viremia. Censored clinical data is used to obtain parameter estimates. We demonstrate that using a reduced set of 16 free parameters, obtained by fixing some parameters at their population averages, the model provides reasonable fits to the patient data and, moreover, that it exhibits good predictive capability. We further show that parameter values obtained for most clinical patients do not admit multiple locally a.s off-treatment equilibria. This suggests that treatment to move from a high viral load equilibrium state to an equilibrium state with a lower (or zero) viral load is not possible for these patients.