Exploration of machine learning techniques in predicting multiple sclerosis disease course.

OBJECTIVE: To explore the value of machine learning methods for predicting multiple sclerosis disease course. METHODS: 1693 CLIMB study patients were classified as increased EDSS≥1.5 (worsening) or not (non-worsening) at up to five years after baseline visit. Support vector machines (SVM) were used to build the classifier, and compared to logistic regression (LR) using demographic, clinical and MRI data obtained at years one and two to predict EDSS at five years follow-up. RESULTS: Baseline data alone provided little predictive value. Clinical observation for one year improved overall SVM sensitivity to 62% and specificity to 65% in predicting worsening cases. The addition of one year MRI data improved sensitivity to 71% and specificity to 68%. Use of non-uniform misclassification costs in the SVM model, weighting towards increased sensitivity, improved predictions (up to 86%). Sensitivity, specificity, and overall accuracy improved minimally with additional follow-up data. Predictions improved within specific groups defined by baseline EDSS. LR performed more poorly than SVM in most cases. Race, family history of MS, and brain parenchymal fraction, ranked highly as predictors of the non-worsening group. Brain T2 lesion volume ranked highly as predictive of the worsening group. INTERPRETATION: SVM incorporating short-term clinical and brain MRI data, class imbalance corrective measures, and classification costs may be a promising means to predict MS disease course, and for selection of patients suitable for more aggressive treatment regimens.

Removing confounding factors via constraint-based clustering: An application to finding homogeneous groups of multiple sclerosis patients.

OBJECTIVES: Confounding factors in unsupervised data can lead to undesirable clustering results. For example in medical datasets, age is often a confounding factor in tests designed to judge the severity of a patient's disease through measures of mobility, eyesight and hearing. In such cases, removing age from each instance will not remove its effect from the data as other features will be correlated with age. Motivated by the need to find homogeneous groups of multiple sclerosis (MS) patients, we apply our approach to remove physician subjectivity from patient data. METHODS: We present a method based on constraint-based clustering to remove the impact of such confounding factors. Given knowledge about which feature (or set of features) is a confounding factor, call it F. Our method first partitions the data into b bins: if F is categorical, instances from the same category construct one bin; if F is numeric, then we split bins such that each bin contains instances of similar F value. Thus each instance is assigned to a single bin for factor F. We then remove feature F from each instance for the remaining steps. Next, we cluster the data separately in each bin. Using these clustering results, we generate pair-wise constraints and then run a constraint-based clustering algorithm to produce a final grouping. RESULTS: In a series of experiments with synthetic datasets, we compare our proposed methods to detrending when one has numeric confounding factors. We apply our method to the Comprehensive Longitudinal Investigation of Multiple Sclerosis at Brigham and Womens Hospital dataset, and find a novel grouping of patients that can help uncover the factors that impact disease progression in MS. CONCLUSIONS: Our method groups data removing the effect of confounding factors without making any assumptions about the form of the influence of these factors on the other features. We identified clusters of MS patients that have clinically recognizable differences. Because patients more likely to progress are found using this approach, our results have the potential to aid physicians in tailoring treatment decisions for MS patients.

Cortical feature analysis and machine learning improves detection of "MRI-negative" focal cortical dysplasia.

Focal cortical dysplasia (FCD) is the most common cause of pediatric epilepsy and the third most common lesion in adults with treatment-resistant epilepsy. Advances in MRI have revolutionized the diagnosis of FCD, resulting in higher success rates for resective epilepsy surgery. However, many patients with histologically confirmed FCD have normal presurgical MRI studies ('MRI-negative'), making presurgical diagnosis difficult. The purpose of this study was to test whether a novel MRI postprocessing method successfully detects histopathologically verified FCD in a sample of patients without visually appreciable lesions. We applied an automated quantitative morphometry approach which computed five surface-based MRI features and combined them in a machine learning model to classify lesional and nonlesional vertices. Accuracy was defined by classifying contiguous vertices as "lesional" when they fell within the surgical resection region. Our multivariate method correctly detected the lesion in 6 of 7 MRI-positive patients, which is comparable with the detection rates that have been reported in univariate vertex-based morphometry studies. More significantly, in patients that were MRI-negative, machine learning correctly identified 14 out of 24 FCD lesions (58%). This was achieved after separating abnormal thickness and thinness into distinct classifiers, as well as separating sulcal and gyral regions. Results demonstrate that MRI-negative images contain sufficient information to aid in the in vivo detection of visually elusive FCD lesions.

Clinical vestibular testing assessed with machine-learning algorithms.

IMPORTANCE: Dizziness and imbalance are common clinical problems, and accurate diagnosis depends on determining whether damage is localized to the peripheral vestibular system. Vestibular testing guides this determination, but the accuracy of the different tests is not known. OBJECTIVE: To determine how well each element of the vestibular test battery segregates patients with normal peripheral vestibular function from those with unilateral reductions in vestibular function. DESIGN, SETTING, AND PARTICIPANTS: Retrospective analysis of vestibular test batteries in 8080 patients. Clinical medical records were reviewed for a subset of individuals with the reviewers blinded to the vestibular test data. INTERVENTIONS: A group of machine-learning classifiers were trained using vestibular test data from persons who were "manually" labeled as having normal vestibular function or unilateral vestibular damage based on a review of their medical records. The optimal trained classifier was then used to categorize patients whose diagnoses were unknown, allowing us to determine the information content of each element of the vestibular test battery. MAIN OUTCOMES AND MEASURES: The information provided by each element of the vestibular test battery to segregate individuals with normal vestibular function from those with unilateral vestibular damage. RESULTS: The time constant calculated from the rotational test ranked first in information content, and measures that were related physiologically to the rotational time constant were 10 of the top 12 highest-ranked variables. The caloric canal paresis ranked eighth, and the other elements of the test battery provided minimal additional information. The sensitivity of the rotational time constant was 77.2%, and the sensitivity of the caloric canal paresis was 59.6%; the specificity of the rotational time constant was 89.0%, and the specificity of the caloric canal paresis was 64.9%. The diagnostic accuracy of the vestibular test battery increased from 72.4% to 93.4% when the data were analyzed with the optimal machine-learning classifier. CONCLUSIONS AND RELEVANCE: Rotational testing should be considered the primary test to diagnose unilateral peripheral vestibular damage in patients with dizziness or imbalance. Most physicians, however, continue to rely on caloric tests to guide their diagnoses. Our results support a significant shift in the approach used to determine diagnoses in patients with vestibular symptoms.

Toward modernizing the systematic review pipeline in genetics: efficient updating via data mining.

PURPOSE: The aim of this study was to demonstrate that modern data mining tools can be used as one step in reducing the labor necessary to produce and maintain systematic reviews. METHODS: We used four continuously updated, manually curated resources that summarize MEDLINE-indexed articles in entire fields using systematic review methods (PDGene, AlzGene, and SzGene for genetic determinants of Parkinson disease, Alzheimer disease, and schizophrenia, respectively; and the Tufts Cost-Effectiveness Analysis (CEA) Registry for cost-effectiveness analyses). In each data set, we trained a classification model on citations screened up until 2009. We then evaluated the ability of the model to classify citations published in 2010 as "relevant" or "irrelevant" using human screening as the gold standard. RESULTS: Classification models did not miss any of the 104, 65, and 179 eligible citations in PDGene, AlzGene, and SzGene, respectively, and missed only 1 of 79 in the CEA Registry (100% sensitivity for the first three and 99% for the fourth). The respective specificities were 90, 93, 90, and 73%. Had the semiautomated system been used in 2010, a human would have needed to read only 605/5,616 citations to update the PDGene registry (11%) and 555/7,298 (8%), 717/5,381 (13%), and 334/1,015 (33%) for the other three databases. CONCLUSION: Data mining methodologies can reduce the burden of updating systematic reviews, without missing more papers than humans.

Single cell time-resolved quorum responses reveal dependence on cell density and configuration.

Bacterial communication via quorum sensing has been extensively investigated in recent years. Bacteria communicate in a complex manner through the production, release, and reception of diffusible low molecular weight chemical signaling molecules. Much work has focused on understanding the basic mechanisms of quorum sensing. As more and more bacteria grow resistant to conventional antibiotics, the development of drugs that do not kill bacteria but instead interrupt their communication is of increasing interest. This study presents a method for analyzing bacterial communication by investigating single cell responses. Most conventional analysis methods for bacterial communication are based on the averaged response from many bacteria, masking how individual cells respond to their immediate environment. We applied a fiber-optic microarray to record cellular communication from single cells. Single cell quorum sensing systems have previously been employed, but the highly ordered array reported here is an improvement because it allows us to simultaneously investigate cellular communication in many different environments with known cellular densities and configurations. We employed this method to detect how genes under quorum regulation are induced or repressed over time on the single cell level and to determine whether cellular density and configuration are indicative of the single cell temporal patterns of gene expression.

Automated storage and retrieval of thin-section CT images to assist diagnosis: system description and preliminary assessment.

A software system and database for computer-aided diagnosis with thin-section computed tomographic (CT) images of the chest was designed and implemented. When presented with an unknown query image, the system uses pattern recognition to retrieve visually similar images with known diagnoses from the database. A preliminary validation trial was conducted with 11 volunteers who were asked to select the best diagnosis for a series of test images, with and without software assistance. The percentage of correct answers increased from 29% to 62% with computer assistance. This finding suggests that this system may be useful for computer-assisted diagnosis.