Clinical and radiological septic joint analysis of spontaneous sternoclavicular joint infections: achieving the best outcomes-a systems engineering approach.

OBJECTIVES: Spontaneous sternoclavicular joint infection (SSCJI) is a rare and poorly understood disease process. This study aims to identify factors guiding effective management strategies for SSCJI by using data mining. METHODS: An Institutional Review Board-approved retrospective review of patients from 2 large hospitals (2010-2022) was conducted. SSCJI is defined as a joint infection without direct trauma or radiation, direct instrumentation or contiguous spread. An interdisciplinary team consisting of thoracic surgeons, radiologists, infectious disease specialists, orthopaedic surgeons, hospital information experts and systems engineers selected relevant variables. Small set data mining algorithms, utilizing systems engineering, were employed to assess the impact of variables on patient outcomes. RESULTS: A total of 73 variables were chosen and 54 analysed against 11 different outcomes. Forty-seven patients [mean age 51 (22-82); 77% male] met criteria. Among them, 34 underwent early joint surgical resection (<14 days), 5 patients received delayed surgical intervention (>14 days) and 8 had antibiotic-only management. The antibiotic-only group had comparable outcomes. Indicators of poor outcomes were soft tissue fluid >4.5 cm, previous SSCJI, moderate/significant bony fragments, HgbA1c >13.9% and moderate/significant bony sclerosis. CONCLUSIONS: This study suggests that targeted antibiotic-only therapy should be considered initially for SSCJI cases while concurrently managing comorbidities. Patients displaying indicators of poor outcomes or no symptomatic improvement after antibiotic-only therapy should be considered for surgical joint resection.

A data-mining approach to predict influent quality.

In wastewater treatment plants, predicting influent water quality is important for energy management. The influent water quality is measured by metrics such as carbonaceous biochemical oxygen demand (CBOD), potential of hydrogen, and total suspended solid. In this paper, a data-driven approach for time-ahead prediction of CBOD is presented. Due to limitations in the industrial data acquisition system, CBOD is not recorded at regular time intervals, which causes gaps in the time-series data. Numerous experiments have been performed to approximate the functional relationship between the input and output parameters and thereby fill in the missing CBOD data. Models incorporating seasonality effects are investigated. Four data-mining algorithms-multilayered perceptron, classification and regression tree, multivariate adaptive regression spline, and random forest-are employed to construct prediction models with the maximum prediction horizon of 5 days.

A data-driven model for maximization of methane production in a wastewater treatment plant.

A data-driven approach for maximization of methane production in a wastewater treatment plant is presented. Industrial data collected on a daily basis was used to build the model. Temperature, total solids, volatile solids, detention time and pH value were selected as parameters for the model construction. First, a prediction model of methane production was built by a multi-layer perceptron neural network. Then a particle swarm optimization algorithm was used to maximize methane production based on the model developed in this research. The model resulted in a 5.5% increase in methane production.

Cancer gene search with data-mining and genetic algorithms.

Cancer leads to approximately 25% of all mortalities, making it the second leading cause of death in the United States. Early and accurate detection of cancer is critical to the well being of patients. Analysis of gene expression data leads to cancer identification and classification, which will facilitate proper treatment selection and drug development. Gene expression data sets for ovarian, prostate, and lung cancer were analyzed in this research. An integrated gene-search algorithm for genetic expression data analysis was proposed. This integrated algorithm involves a genetic algorithm and correlation-based heuristics for data preprocessing (on partitioned data sets) and data mining (decision tree and support vector machines algorithms) for making predictions. Knowledge derived by the proposed algorithm has high classification accuracy with the ability to identify the most significant genes. Bagging and stacking algorithms were applied to further enhance the classification accuracy. The results were compared with that reported in the literature. Mapping of genotype information to the phenotype parameters will ultimately reduce the cost and complexity of cancer detection and classification.

Patient-recognition data-mining model for BCG-plus interferon immunotherapy bladder cancer treatment.

Bladder cancer is the fifth most common malignant disease in the United States with an annual incidence of around 63,210 new cases and 13,180 deaths. The cost for providing care for patients with bladder cancer disease is high. Bladder cancer treatment options such as immunotherapy, chemotherapy, radiation therapy, transurethral resection, and cystectomy, are used with varying success rates. In this research, data from a nationwide bacillus Calmette-Gue rin (BCG) plus interferon-alpha (IFN-alpha) immunotherapy clinical trial was considered. Data mining algorithms were used to analyze the effectiveness of immunotherapy treatment and to understand the prominent parameters and their interactions. The extracted knowledge was used to build a patient recognition model for prediction of treatment outcomes. The data was analyzed to understand the impact of various parameters on the treatment outcome. A list of significant parameters such as cumulative tumor size, presence of residual disease, stages of prior bladder cancer, current state of bladder cancer, and the presence of current bladder cancer (T1) is provided. The decision-making approach outlined in the paper supplemented with additional knowledge bases will lead to a comprehensive analytical road map of the BCG/IFN-alpha immunotherapy treatment. It will provide individualized guidelines for each stage of the treatment as well as measure the success of the treatment.

Hypoplastic left heart syndrome: knowledge discovery with a data mining approach.

Hypoplastic left heart syndrome (HLHS) affects infants and is uniformly fatal without surgical palliation. Post-surgery mortality rates are highly variable and dependent on postoperative management. A data acquisition system was developed for collection of 73 physiologic, laboratory, and nurse-assessed parameters. The acquisition system was designed for the collection on numerous patients. Data records were created at 30s intervals. An expert-validated wellness score was computed for each data record. To efficiently analyze the data, a new metric for assessment of data utility, the combined classification quality measure, was developed. This measure assesses the impact of a feature on classification accuracy without performing computationally expensive cross-validation. The proposed measure can be also used to derive new features that enhance classification accuracy. The knowledge discovery approach allows for instantaneous prediction of interventions for the patient in an intensive care unit. The discovered knowledge can improve care of complex to manage infants by the development of an intelligent bedside advisory system.

Predicting survival time for kidney dialysis patients: a data mining approach.

The cost for providing care for patients on hemodialysis due to end stage kidney disease is high. Finding ways to improve patient outcomes and reduce the cost of dialysis is important. Dialysis care is intricate and multiple factors may influence patient survival. Over 50 parameters may be monitored on a regular basis in providing kidney dialysis treatments. Understanding the collective role of these parameters in determining outcomes for an individual patient and administering individualized treatments allowing specific interventions is a challenge. Individual patient survival may depend on a complex interrelationship between multiple demographic and clinical parameters, medications, medical interventions, and the dialysis treatment prescription. In this research, data preprocessing, data transformations, and a data mining approach are used to elicit knowledge about the interaction between many of these measured parameters and patient survival. Two different data mining algorithms were employed for extracting knowledge in the form of decision rules. These rules were used by a decision-making algorithm, which predicts survival of new unseen patients. Important parameters identified by data mining are interpreted for their medical significance. The concepts introduced in this research have been applied and tested using data collected at four dialysis sites. The computational results are reported in the paper.

Data mining and genetic algorithm based gene/SNP selection.

OBJECTIVE: Genomic studies provide large volumes of data with the number of single nucleotide polymorphisms (SNPs) ranging into thousands. The analysis of SNPs permits determining relationships between genotypic and phenotypic information as well as the identification of SNPs related to a disease. The growing wealth of information and advances in biology call for the development of approaches for discovery of new knowledge. One such area is the identification of gene/SNP patterns impacting cure/drug development for various diseases. METHODS: A new approach for predicting drug effectiveness is presented. The approach is based on data mining and genetic algorithms. A global search mechanism, weighted decision tree, decision-tree-based wrapper, a correlation-based heuristic, and the identification of intersecting feature sets are employed for selecting significant genes. RESULTS: The feature selection approach has resulted in 85% reduction of number of features. The relative increase in cross-validation accuracy and specificity for the significant gene/SNP set was 10% and 3.2%, respectively. CONCLUSION: The feature selection approach was successfully applied to data sets for drug and placebo subjects. The number of features has been significantly reduced while the quality of knowledge was enhanced. The feature set intersection approach provided the most significant genes/SNPs. The results reported in the paper discuss associations among SNPs resulting in patient-specific treatment protocols.

Relabeling algorithm for retrieval of noisy instances and improving prediction quality.

A relabeling algorithm for retrieval of noisy instances with binary outcomes is presented. The relabeling algorithm iteratively retrieves, selects, and re-labels data instances (i.e., transforms a decision space) to improve prediction quality. It emphasizes knowledge generalization and confidence rather than classification accuracy. A confidence index incorporating classification accuracy, prediction error, impurities in the relabeled dataset, and cluster purities was designed. The proposed approach is illustrated with a binary outcome dataset and was successfully tested on the standard benchmark four UCI repository dataset as well as bladder cancer immunotherapy data. A subset of the most stable instances (i.e., 7% to 51% of the sample) with high confidence (i.e., between 64%-99.44%) was identified for each application along with most noisy instances. The domain experts and the extracted knowledge validated the relabeled instances and corresponding confidence indexes. The relabeling algorithm with some modifications can be applied to other medical, industrial, and service domains.

Multiobjective optimization of temporal processes.

This paper presents a dynamic predictive-optimization framework of a nonlinear temporal process. Data-mining (DM) and evolutionary strategy algorithms are integrated in the framework for solving the optimization model. DM algorithms learn dynamic equations from the process data. An evolutionary strategy algorithm is then applied to solve the optimization problem guided by the knowledge extracted by the DM algorithm. The concept presented in this paper is illustrated with the data from a power plant, where the goal is to maximize the boiler efficiency and minimize the limestone consumption. This multiobjective optimization problem can be either transformed into a single-objective optimization problem through preference aggregation approaches or into a Pareto-optimal optimization problem. The computational results have shown the effectiveness of the proposed optimization framework.