Receiver Operating Characteristic Graph Research Articles

The Optimal Variation Range of Posterior Condylar Offset Associated with Positive Clinical Outcomes of Primary Total Knee Arthroplasty: A Retrospective Analysis.

Posterior condylar offset (PCO) and anterior condylar offset (ACO) exert an influence on the sagittal alignment in total knee arthroplasty (TKA). However, there is no common consensus that the variation range of posterior condylar offset (PCO) is associated with patient-reported outcome measures (PROMs) and the optimum variation range of PCO. This study aims to investigate the correlation between PCO and the PROMs of primary TKA for osteoarthritis (OA) and find out the optimal variation range of the PCO. In this study, we performed a radiographic analysis of 106 patients (112 knees) with primary TKA. Patients were divided into two cohorts (A and B) according to the Western Ontario and McMaster Universities Osteoarthritis index (WOMAC). Correlations between the sagittal parameter and WOMAC were investigated using univariate and multivariate analysis. The receiver operating characteristic (ROC) curve was used to establish the cut-off value for the optimal variation range. We then further investigated how different variation range affects the WOMAC subscale score and forgotten-joint-score-12 (FJS-12). Univariate analysis revealed a correlation between the variation range of PCO (p < 0.01), ACO (p < 0.01) and PROMs. Multivariate analysis showed that only PCO was associated with PROMs. In the ROC graph, the cut-off value of the variation range of PCO is 2.85 mm (AUC = 0.66, Youden index = 0.26). The WOMAC functional ability score of the group outside the PCO variation range of 2.85 mm significantly increased compared to the group within the range. In this study, PCO variation was significantly associated with clinical outcomes in TKA and the optimal PCO variation range was within 2.85 mm. Maintaining the PCO variation within 2.85 mm could enhance functional recovery and patient satisfaction.

Optimizing biocompatible materials for personalized medical implants using reinforcement learning and Bayesian strategies

This study presents a comprehensive research methodology integrating computational approaches, statistical analysis, and visualization techniques to predict biocompatible materials for medical implants and evaluate predictive model performance (Whiting K. 2020, October). The initial phase involves data acquisition and preprocessing, organizing a representative dataset into a pandas data frame. Visualization of the dataset through bar, pie, and line charts provides insights into relationships between materials and functional attributes. The subsequent phase focuses on evaluating a predictive model using simulated datasets and key metrics such as accuracy, precision, recall, F1 score, and the receiver operating characteristics (ROC) curve with an area under the curve (AUC) value. Performance metrics are visually represented through bar charts and ROC curves, aiding stakeholders in understanding the model’s strengths and areas for improvement. The confusion matrix offers a granular examination of the model’s classification performance. The results and discussion section delves into graphical representations, emphasizing the material vs. strength/conductance/resistance/function chart, elucidating the diverse functional profiles of materials. The distribution of material functionality pie chart succinctly illustrates the proportional contribution of each material, aiding informed decision-making in material selection. The materials performance graph provides a nuanced understanding of material characteristics, guiding the development of personalized healthcare solutions. Model performance metrics and receiver operating characteristics graphs comprehensively assess the predictive model, while the confusion matrix details classification outcomes. This methodology and its visualizations contribute to predicting biocompatible materials, emphasizing the significance of advanced computational approaches for efficiently navigating the complex material space. The study’s outcomes inform both material scientists and healthcare professionals, guiding the development of personalized healthcare solutions tailored to specific patient needs.

Receiver Operating Characteristic Optimization Based on Convex Hull and Evolutionary Algorithm

Classification is basically a Multi-objective problem. The efficiency of classification vastly depends on performance of a classifier which can be evaluated on the basis of Receiver Operating Characteristic (ROC) graph, Area under Curve (AUC), and selection of different threshold values are generally used as a tool. In machine learning, generally, 2-D classifiers are available that deal with bi-objective problems where overlapping of class may occur i.e. sensitivity and specificity may overlap. Recently, multi-class classification in which classes are mutually exclusive is in research trends along with the evolutionary algorithm. The application of ROC graph is extended to evaluate multi-dimensional classification as it is cost sensitive learning. The goal of this paper is to gather recent achievements in the field at one place to analyze the classifier performance for multi-dimensional classification problems using convex hull and evolutionary algorithms. In this paper, we tried to cover all the existing recent advance techniques in maximization of ROC and proposed a convex hull and evolutionary algorithm based new model for ROC maximization.

Research Note: Is infrared thermography an appropriate method for early detection and objective quantification of plumage damage in white and brown feathered laying hens?

For the standardized assessment of plumage damage in laying hens, imaging techniques can be used in addition to visual plumage scoring (PS). In this study, the diagnostic accuracy of infrared thermography (IRT) was analyzed in white-feathered (WL) and brown-feathered laying hens (BL) with PS as a reference. In 28 flocks, a 3-level PS and IRT were performed 8 times for the dorsal neck, back, and belly plumage. A total of 3,600 records for WL and 7,600 records for BL were available for both PS and IRT in each of the 3 body regions. Receiver operating characteristic (ROC) analyses were used to investigate diagnostic accuracy. High-accuracy detection was found for severe plumage damage on the back (WL—sensitivity: 96.0%, specificity: 98.3%; BL—sensitivity: 96.1%, specificity: 98.9%) and belly plumage (WL—sensitivity: 96.3%, specificity: 95.7%; BL—sensitivity: 95.3%, specificity: 97.2%), but insufficient accuracy for distinguishing between intact plumage and moderate plumage damage in all 3 body regions (sensitivity: 31.7–71.5%; specificity: 70.4–98.1%). The area under the curve (AUC) of the ROC graphs differed significantly between BL and WL for the back and belly plumage (P ≤ 0.05). We concluded that IRT is a suitable tool to objectively detect severe plumage damage but not for early detection of incipient plumage loss.

The Utility of Receiver Operating Characteristic Curve in Educational Assessment: Performance Prediction

When examinees are classified into groups based on scores from educational assessment, two indices are widely used to gauge the psychometric quality of the classifications: accuracy and consistency. The two indices take correct classifications into consideration while overlooking incorrect ones, where unbalanced class distribution threatens the validity of results from the accuracy and consistency indices. The single values produced from the two indices also fail to address the inconsistent accuracy of the classifier across different cut score locations. The current study proposed the concept of classification quality, which utilizes the receiver operating characteristics (ROC) graph to comprehensively evaluate the performance of classifiers. The ROC graph illustrates the tradeoff between benefits (true positive rate) and costs (false positive rate) in classification. In this article, a simulation study was conducted to demonstrate how to generate and interpret ROC graphs in educational assessment and the benefits of using ROC graphs to interpret classification quality. The results show that ROC graphs provide an efficient approach to (a) visualize the fluctuating performance of scoring classifiers, (b) address the unbalanced class distribution issue inherent in the accuracy and consistency indices, and (c) produce more accurate estimation of the classification results.

A template-free machine vision-based crop row detection algorithm

Due to the increase in the use of precision agriculture, field trials have increased in size to allow for genomic selection tool development by linking quantitative phenotypic traits to sequence variations in the DNA of various crops. Crop row detection is an important step to enable the development of an efficient downstream analysis pipeline for genomic selection. In this paper, an efficient crop row detection algorithm was proposed that detected crop rows in colour images without the use of templates and most other pre-information such as number of rows and spacing between rows. The method only requires input on field weed intensity. The algorithm was robust in challenging field trial conditions such as variable light, sudden shadows, poor illumination, presence of weeds and noise and irregular crop shape. The algorithm can be applied to crop images taken from the top and side views. The algorithm was tested on a public dataset with side view images of crop rows and on Genomic Sub-Selection dataset in which images were taken from the top view. Different analyses were performed to check the robustness of the algorithm and to the best of authors’ knowledge, the Receiver Operating Characteristic graph has been applied for the first time in crop row detection algorithm testing. Lastly, comparing this algorithm with several state-of-the-art methods, it exhibited superior performance.

Robust image hashing with compressed sensing and ordinal measures

Image hashing is an efficient technology for processing digital images and has been successfully used in image copy detection, image retrieval, image authentication, image quality assessment, and so on. In this paper, we design a new image hashing with compressed sensing (CS) and ordinal measures. This hashing method uses a visual attention model called Itti model and Canny operator to construct an image representation, and exploits CS to extract compact features from the representation. Finally, the CS-based compact features are quantized via ordinal measures. L2 norm is used to judge similarity of hashes produced by the proposed hashing method. Experiments about robustness validation, discrimination test, block size discussion, selection of visual attention model, selection of quantization scheme, and effectiveness of the use of ordinal measures are conducted to verify performances of the proposed hashing method. Comparisons with some state-of-the-art algorithms are also carried out. The results illustrate that the proposed hashing method outperforms some compared algorithms in classification according to ROC (receiver operating characteristic) graph.

Forecasting seismic activity rates in northwest Himalaya through multivariate autoregressive forecast of seismicity algorithm

In this study, a model based on multivariate autoregressive forecast of seismicity (MARFS) algorithm is adopted to forecast seismic activity rates in northwest Himalaya, using the compiled homogenized moment magnitude (MW) based catalogue. For this purpose, each source zone delineated by Yadav et al. (Pure Appl Geophys 170:283–295, 2012) is divided into a spatial grid interval of 0.5° × 0.5° while the entire catalogue span (1975–2010) is segregated into six time periods/grids to estimate seismic activity rates spatially and temporally. These seismic activity rates which are estimated from spatial density map of hypocenters exhibit high values in Chaman Fault (Zone 1), Hindukush-Pamir region (Zone 3) and the mega thrust systems, i.e., Main Central Thrust, Main Boundary Thrust and Himalayan Frontal Thrust (Zone 4). Then, the seismic activity rates during 2011–2016 could be forecasted by extrapolating (through auto-regression procedure) those observed for previous time periods. The forecast seismic activity rates are estimated within the values of 0 and 7.57 with high values primarily observed in Hindukush-Pamir region of Zone 3 and the gently north-dipping thrust fault systems (Main Central Thrust, Main Boundary Thrust, Himalayan Frontal Thrust) of Zone 4. Finally, the associated area under the curve of receiver operating characteristics graph suggests the superiority of forecasting model with respect to random prediction, whereas results of the data-consistency test, i.e., N test of our model, exhibit consistency in between the observed and simulated likelihoods. Moreover, the hypothetical t test performed in between the spatial grids of forecast seismic activity rates and observed seismic activity rates confirms that the former is consistent with the latter.

Performance evaluation of optical sorting in mineral processing – A case study with quartz, magnesite, hematite, lignite, copper and gold ores

Optical sorting is increasingly playing an important role in mineral processing. Therefore, a better understanding of this method is required concerning general properties and mineral sorting applications. To date, optical sorting has been widely studied in terms of industrial applications and performance evaluation particularly in mineral processing. Nevertheless, process optimization requires better understanding of qualitative and quantitative figures based on real life sorting applications.In this study, the relationship between feed rate and separation efficiency of a gravity type and visible light responsive sensor mounted optical sorter (VIS sorter) was investigated. In addition to the feed rate, the effect of particle size on separation performance was also discussed. Laboratory studies included a selection of material, sample preparation, and optical sorting tests with magnesite, quartz, lignite, hematite, copper and gold ore samples. Experimental studies performed with different size fractions at varying feed rates were followed by evaluation of results with ROC (receiver operating characteristics) graphs. Recovery, grade, and capacity indicators are important to define the efficiency of the optical sorting. Alternatively, ROCs highlight missing particles and false alarm rates, which are used to characterize set points and detectability of the equipment/operation. In the tests performed with quartz and magnesite samples, % weights of valuable (white) and non–valuable (colored) particles in products were directly used to express the performance of optical sorting with ROC. Additionally, % Fe content, % dry ash content, % copper recovery and % gold recovery figures were used to define the ROC parameters for hematite, lignite, copper and gold ore samples.Results showed that the performance of the VIS type optical sorter mainly depends on correctly sorted amounts of valuable and non–valuable particles. Alongside with the evaluation of change in sorting performances in accordance with feed rate, ore type, and particle size; pre-concentration of copper and gold ores, concentration of hematite from alkaline waste, dry cleaning of lignite and market quality magnesite and quartz sorting applications with VIS type optical sorter were also discussed.

Time-frequency analysis and support vector machine in automatic detection of defect from vibration signal of centrifugal pump

Centrifugal pumps operate at moderate to high speed. Contamination in the fluid in terms of solid particles and chemically reactive substances causes damage to the impeller, casing, and seals. A defect in bearing due to improper lubrication, adverse loading and manufacturing defect may also affect the performance of the pump. Hence there is a need to develop a reliable procedure for defect identification in the centrifugal pump. A robust automated signal processing algorithm is proposed for the purpose. Features sensitive to defective conditions are extracted from raw signal and scale marginal integration graph. The genetic algorithm (GA) is used to find the optimal parameters of support vector machine (SVM). Using the optimal parameters, training of SVM is carried out for the learning of defective conditions of the pump. After training, features are applied to SVM for the identification of the defective condition of the pump. The performance evaluation of the proposed method is made using receiver operating characteristics graph and is found to be reliable. The overall recognition rate of the proposed method in identifying the specific conditions of the pump is 96.66%. In this work, an attempt is also been made to reduce the training time of GA-SVM model.

Somatic cell count thresholds in composite and quarter milk samples as indicator of bovine intramammary infection status.

The objective of the study was to establish an operational somatic cell count (SCC) threshold to predict the presence of intramammary infection (IMI) in composite milk samples and compare findings with those in quarter milk samples. South African dairy producers now preferred composite milk samples for herd udder health analysis because of increasing cow numbers, convenience of sampling and lower cost. A retrospective study was conducted on 345 461 composite and 89 638 quarter milk samples from South African herds. Variance estimates for the proportion of quarter samples testing positive were adjusted to account for the lack of their independence within individual cows. The IMI at SCC thresholds of 150 000 cells/mL and 200 000 cells/mL differed only by 3.26% in composite milk samples. Youden’s index indicated the optimum SCC thresholds for composite and quarter milk samples as 150 000 cells/mL and 200 000 cells/mL, respectively. At 150 000 cells/mL, sensitivity (95% confidence intervals [CI]) in composite milk samples was 65.3% (64.0%, 66.6%) and specificity was 66.8% (65.7%, 67.9%); and in quarter milk samples, sensitivity at 200 000 cells/mL was 70.8% (69.5%, 72.0%) and specificity was 63.6% (62.4%, 64.8%). The likelihood of infection for udders and quarters, respectively, was 1.034 and 1.327 at an SCC threshold of 150 000 cells/mL and 0.864 cells/mL and 1.177 cells/mL at 200 000 cells/mL. The area under the curve of the receiver operating characteristics graph was 0.7084 and 0.7277 for composite and quarter samples, respectively, indicating that the SCC test could be considered as a good indicator of IMI in both sample types.

Features of the Area under the Receiver Operating Characteristic (ROC) Curve. A Good Practice

The area under the receiver operating characteristic (ROC) curve is a measure of discrimination ability used in diagnostic and prognostic research. The ROC plot is usually represented without additional information about decision thresholds used to generate the graph. In our article, we show that adding at least one or more informative cutoff points on the ROC graph facilitates the characterization of the test and the evaluation of the discriminatory capacities, which can result in more informed medical decisions. We use the rocreg and rocregplot commands.

Development of a method for monitoring of insect induced forest defoliation – limitation of MODIS data in Fennoscandian forest landscapes

We investigated if coarse-resolution satellite data from the MODIS sensor can be used for regional monitoring of insect disturbances in Fennoscandia. A damage detection method based on z-scores of seasonal maximums of the 2-band Enhanced Vegetation Index (EVI2) was developed. Time-series smoothing was applied and Receiver Operating Characteristics graphs were used for optimisation. The method was developed in fragmented and heavily managed forests in eastern Finland dominated by Scots pine ( L.) (pinaceae) and with defoliation of European pine sawfly ( Geoffr.) (Hymenoptera: Diprionidae) and common pine sawfly ( L.) (Hymenoptera: Diprionidae). The method was also applied to subalpine mountain birch ( N.I. Orlova) forests in northern Sweden, infested by autumnal moth (Borkhausen) and winter moth ( L.). In Finland, detection accuracies were fairly low with 50% of the damaged stands detected, and a misclassification of healthy stands of 22%. In areas with long outbreak histories the method resulted in extensive misclassification. In northern Sweden accuracies were higher, with 75% of the damage detected and a misclassification of healthy samples of 19%. Our results indicate that MODIS data may fail to detect damage in fragmented forests, particularly when the damage history is long. Therefore, regional studies based on these data may underestimate defoliation. However, the method yielded accurate results in homogeneous forest ecosystems and when long-enough periods without damage could be identified. Furthermore, the method is likely to be useful for insect disturbance detection using future medium-resolution data, e.g. from Sentinel‑2.Pinus sylvestrisNeodiprion sertiferDiprion piniBetula pubescensssp. Czerepanovii Epirrita autumnata Operophtera brumata

Abstract 14854: Optimal Cut-off Points of Waist Circumference for Detecting Central Obesity and Severe Central Obesity in Chinese Adults

Introduction: Given that no consensus has yet been reached over the optimal cut-off points of waist circumference (WC) for Chinese, this study aimed to determine the appropriate cut-off points of WC for detecting central obesity and severe central obesity in a Chinese adult population. Methods: Data from the cross-sectional survey of the PRC-USA Collaborative study of Cardiovascular and Cardiopulmonary Epidemiology in 1993-1994 was used, including 10265 subjects (4921 men and 5344 women) aged 35-69 years. Each integer of WC in centimeters was used as the cut-off point to detect clustering of cardiovascular risk factors, which was defined as the presence of two or more risk factors among hypertension, diabetes, hypercholesterolemia, hypertriglyceridemia and low levels of high-density lipoprotein cholesterol. Based on the Receiver operating characteristic (ROC) curve analysis, the WC value corresponding to the point on the ROC curve nearest to the upper left corner of the ROC graph was considered the optimal cut-off for central obesity and the value corresponding to the point with specificity of 90% or more was considered the optimal cut-off for severe central obesity. Results: The optimal cut-off value of WC to detect clustering of risk factors was ≥84 cm for men and was ≥80 cm for women with the shortest distance to the upper left corner being 0.4304 and 0.4504, respectively. The cut-off values of WC with specificity above 90% were ≥93 cm and ≥91 cm for men and women respectively. Conclusions: These results were equal or similar to the WC cut-off points proposed by the Guidelines for Prevention and Control of Overweight and Obesity in Chinese Adults, i.e., ≥85/80 cm (men/women) for increased risk for obesity related diseases and ≥95/90 cm (men/women) for much higher risk. For practical reasons, WC≥85/80 cm and ≥95/90 cm (men/women) could be the optimal cut-off values for detecting central obesity and severe central obesity, respectively, in Chinese adult population.

Cut-off values of waist circumference for central obesity in Chinese adults

To investigate the appropriate cut-off values of waist circumference(WC)for central obesity and severe central obesity in Chinese adults. A total of 10 265 participants aged 35-69 years from the cross-sectional survey of the PRC-USA Collaborative Study of Cardiovascular and Cardiopulmonary Epidemiology between 1993 and 1994 with integral data were included. Each integer unit in centimeters of WC in a given range was used as the cut-off point to detect clustering of risk factors, which was defined as an individual with 2 or more risk factors including hypertension, diabetes, hypercholesterolemia, hypertriglyceridemia and low levels of high-density lipoprotein cholesterol. Sensitivity, specificity and distance from the receiver operating characteristic (ROC) curves to the upper left corner of the ROC graph were calculated. The WC value corresponding to the point on the ROC curve nearest to the upper left corner was considered as the optimal cut-off value for central obesity and the value corresponding to the point with specificity of 90% or more was considered as the optimal cut-off for severe central obesity. The mean WC was (80.5 ± 9.9) cm in men and (77.8 ± 10.0) cm in women; 18.1% (890/4 921) of men and 14.5% (776/5 344) of women were identified with two or more major risk factors. Based on the ROC curve analysis, the optimal value of WC to detect clustering of risk factors was ≥ 84 cm for men and ≥ 80 cm for women, and the shortest distance to the upper left corner was 0.430 and 0.450, respectively. The cut-off values of WC to detect clustering of risk factors with specificity of 90% or more were ≥ 93 cm and ≥ 91 cm for men and women, respectively. The cut-off points of WC for central obesity and severe central obesity in Chinese adults obtained from this study are equal or similar to the WC cut-off values proposed by the Guidelines for Prevention and Control of Overweight and Obesity in Chinese Adults.

A method for predicting the factor of safety of an infinite slope based on the depth ratio of the wetting front induced by rainfall infiltration

Abstract. Most landslides in Korea are classified as shallow landslides with an average depth of less than 2 m. These shallow landslides are associated with the advance of a wetting front in the unsaturated soil due to rainfall infiltration, which results in an increase in water content and a reduction in the matric suction in the soil. Therefore, this study presents a modified equation of infinite slope stability analysis based on the concept of the saturation depth ratio to analyze the slope stability change associated with the rainfall on a slope. A rainfall infiltration test in unsaturated soil was performed using a column to develop an understanding of the effect of the saturation depth ratio following rainfall infiltration. The results indicated that the rainfall infiltration velocity due to the increase in rainfall in the soil layer was faster when the rainfall intensity increased. In addition, the rainfall infiltration velocity tends to decrease with increases in the unit weight of soil. The proposed model was applied to assess its feasibility and to develop a regional landslide susceptibility map using a geographic information system (GIS). For that purpose, spatial databases for input parameters were constructed and landslide locations were obtained. In order to validate the proposed approach, the results of the proposed approach were compared with the landslide inventory using a ROC (receiver operating characteristics) graph. In addition, the results of the proposed approach were compared with the previous approach used: a steady-state hydrological model. Consequently, the approach proposed in this study displayed satisfactory performance in classifying landslide susceptibility and showed better performance than the steady-state approach.

A weighted-ROC graph based metric for image segmentation evaluation

Evaluation of image segmentation algorithms is a crucial task in the image processing field. Generally, traditional objective evaluation measures, such as ME and JS, always give the same treatment to the object pixels and the background pixels in images, which is not reasonable in practical applications. To overcome this problem, a new objective evaluation metric based on the weighted-ROC graph is proposed in this paper. Considering that pixels in different positions may gain different importance, each pixel is given a weight based on its spatial information. The ROC (receiver operating characteristic) graph with weighting strategy is constructed to evaluate the performance of segmentation algorithms quantitatively. The proposed metric focuses on the segmented objects, which is similar to human visual system. Meanwhile, it reserves the robustness of ROC against the region imbalance. The experimental results on various images show that the proposed metric gives more reasonable evaluation results than other metrics.

Evaluace binárních klasifikačních úloh v ekonomické predikci

In the area of economical classification tasks, the accuracy maximization is often used to evaluate classifier performance. Accuracy maximization (or error rate minimization) suffers from the assumption of equal false positive and false negative error costs. Furthermore, accuracy is not able to express true classifier performance under skewed class distribution. Due to these limitations, the use of accuracy on real tasks is questionable. In a real binary classification task, the difference between the costs of false positive and false negative error is usually critical. To overcome this issue, the Receiver Ope­rating Characteristic (ROC) method in relation to decision-analytic principles can be used. One essential advantage of this method is the possibility of classifier performance visualization by means of a ROC graph. This paper presents concrete examples of binary classification, where the inadequacy of accuracy as the evaluation metric is shown, and on the same examples the ROC method is applied. From the set of possible classification models, the probabilistic classifier with continuous output is under consideration. Mainly two questions are solved. Firstly, the selection of the best classifier from a set of possible classifiers. For example, accuracy metric rates two classifiers almost equiva­lently (87.7 % and 89.3 %), whereas decision analysis (via costs minimization) or ROC analysis reveal differe­nt performance according to target conditions of unequal error costs of false positives and false negatives. Secondly, the setting of an optimal decision threshold at classifier’s output. For example, accuracy maximization finds the optimal threshold at classifier’s output in value of 0.597, but the optimal threshold respecting higher costs of false negatives is discovered by costs minimization or ROC analysis in a value substantially lower (0.477).

Comparison of Objective Functions for Feed-forward Neural Network Classifiers Using Receiver Operating Characteristics Graph

When developing a classifier using various objective functions, it is important to compare the performances of the classifiers. Although there are statistical analyses of objective functions for classifiers, simulation results can provide us with direct comparison results and in this case, a comparison criterion is considerably critical. A Receiver Operating Characteristics (ROC) graph is a simulation technique for comparing classifiers and selecting a better one based on a performance. In this paper, we adopt the ROC graph to compare classifiers trained by mean-squared error, cross-entropy error, classification figure of merit, and the n-th order extension of cross-entropy error functions. After the training of feed-forward neural networks using the CEDAR database, the ROC graphs are plotted to help us identify which objective function is better.

Advanced Machine Learning for classification of EEG traits as Parkinson’s biomarker

We aim to develop non-­‐invasive, low-­‐cost preclinical markers for synucleinopathies (Parkinson’s Disease -­‐ PD or Dementia with Lewy Bodies -­‐ DLB) with impact on neuroprotection. In a first stage we have applied Machine Learning techniques to the analysis of the spontaneous, waking EEG of REM Sleep Behavior Disorder (RBD) patients. We know that patients with RBD may evolve towards PD and other synucleinopathies (i.e. DLB) providing a useful guide in the search for biomarkers. EEG has proven to be sensitive to brain alterations related to both RBD and PD patients, and it is known that patients with either RBD or de novo PD without obvious cognitive alterations show similar EEG/MEG alterations (mainly slowing of the EEG) while awake. In the work described herein we achieve the classification of patients according to their ultimate diagnosis, differentiating among different patient groups. We have analyzed EEG data from a study of RBD patients at the Center for Advanced Research in Sleep Medicine of Montreal. The EEG was recorded from 8 patients who years later evolved to PD, 8 patients who later evolved to LBD, 10 patients with RBD who did not convert, and 17 healthy controls. It is worth mentioning that 80% of the PD and DLB patients developed disease at a follow-­‐up of 8 years. Support Vector Machines were applied to different feature types. Absolute Band Power features extracted each 4 seconds outperform the other ones. After non-­‐parametric feature selection of the 5 most discriminative channel-­‐band combinations, classification was applied. A procedure for performance estimation close to operational conditions, which is denoted as leave-­‐pair-­‐subjects-­‐ out, has been employed for evaluation of the implemented system. Excellent performance levels, AUC 93.75-­‐99%, were obtained for all realized group comparisons. In conclusion, we believe that these results support the idea that classification of EEG shows great potential as a preclinical biomarker. Our results confirm the existent literature, where EEG slowing characterizes RBD and PD/DLB patients. Furthermore we have attained this characterization at the individual subject level for the first time by employing machine learning. Given that EEG was acquired 8 years before disease development in a majority of cases, we believe that the utilization of the developed system for conversion prognosis of synucleinopathies (PD or DLB) will have real impact on both early treatment and drug development.