Proportion Of Correct Decisions Research Articles

A machine learning model for orthodontic extraction/non-extraction decision in a racially and ethnically diverse patient population

The purpose of the present study was to create a machine learning (ML) algorithm with the ability to predict the extraction/non-extraction decision in a racially and ethnically diverse sample. Data was gathered from the records of 393 patients (200 non-extraction and 193 extraction) from a racially and ethnically diverse population. Four ML models (logistic regression [LR], random forest [RF], support vector machine [SVM], and neural network [NN]) were trained on a training set (70% of samples) and then tested on the remaining samples (30%). The accuracy and precision of the ML model predictions were calculated using the area under the curve (AUC) of the receiver operating characteristics (ROC) curve. The proportion of correct extraction/non-extraction decisions was also calculated. The LR, SVM, and NN models performed best, with an AUC of the ROC of 91.0%, 92.5%, and 92.3%, respectively. The overall proportion of correct decisions was 82%, 76%, 83%, and 81% for the LR, RF, SVM, and NN models, respectively. The features found to be most helpful to the ML algorithms in making their decisions were maxillary crowding/spacing, L1-NB (mm), U1-NA (mm), PFH:AFH, and SN-MP(̊), although many other features contributed significantly. ML models can predict the extraction decision in a racially and ethnically diverse patient population with a high degree of accuracy and precision. Crowding, sagittal, and vertical characteristics all featured prominently in the hierarchy of components most influential to the ML decision-making process.

A Crossover Study Comparing the Effect of a Processed vs. Unprocessed Diet on the Spatial Learning Ability of Zebrafish

The current study aimed to compare the short-term effects of processed versus unprocessed food on spatial learning and survival in zebrafish (Danio rerio). Given public concern regarding processed foods, the results of the current study can inform consumer decision-making. We hypothesized that an unprocessed diet would improve learning and survival. Zebrafish were randomly assigned to a diet of brine shrimp flakes (processed) or live brine shrimp (unprocessed). Spatial learning was evaluated throughout the study by recording fish decisions (correct vs. incorrect) and time taken for decision. Our results show no statistically significant difference in the proportion of correct decisions or in the time taken to make decisions when the groups are compared. Notably, fish receiving unprocessed food had significantly lower mortality than those receiving processed food (p = 0.027). We concluded that while the zebrafish make progressively better and faster decisions, diet type does not contribute to improved learning. However, consumption of unprocessed diet may have survival benefits in stressful environments. Future studies may further analyze this association, utilizing larger sample size and longer study duration to clearly assess the effects of long-term exposure to an unprocessed vs. raw unprocessed diet on learning. While it is an association that needs further evaluation, the current study indicates the potential benefits of an unprocessed diet in coping with stress. Considering that zebrafish are effective models of human cognition, this has implications for human consumption of processed vs. unprocessed foods as well.

Evaluating persistence of shape information using a matching protocol.

Many laboratories have studied persistence of shape information, the goal being to better understand how the visual system mediates recognition of objects. Most have asked for recognition of known shapes, e.g., letters of the alphabet, or recall from an array. Recognition of known shapes requires access to long-term memory, so it is not possible to know whether the experiment is assessing short-term encoding and working memory mechanisms, or has encountered limitations on retrieval from memory stores. Here we have used an inventory of unknown shapes, wherein a string of discrete dots forms the boundary of each shape. Each was displayed as a target only once to a given respondent, with recognition being tested using a matching task. Analysis based on signal detection theory was used to provide an unbiased estimate of the probability of correct decisions about whether comparison shapes matched target shapes. Four experiments were conducted, which found the following: a) Shapes were identified with a high probability of being correct with dot densities ranging from 20% to 4%. Performance dropped only about 10% across this density range. b) Shape identification levels remained very high with up to 500 milliseconds of target and comparison shape separation. c) With one-at-a-time display of target dots, varying the total time for a given display, the proportion of correct decisions dropped only about 10% even with a total display time of 500 milliseconds. d) With display of two complementary target subsets, also varying the total time of each display, there was a dramatic decline of proportion correct that reached chance levels by 500 milliseconds. The greater rate of decline for the two-pulse condition may be due to a mechanism that registers when the number of dots is sufficient to create a shape summary. Once a summary is produced, the temporal window that allows shape information to be added may be more limited.

The use of data mining to assist crop protection decisions on kiwifruit in New Zealand

Data mining algorithms were used to develop models to forecast the outcome of leafroller pest monitoring decisions on ‘Hayward’ kiwifruit crops in New Zealand. Using industry spray diary and pest monitoring data gathered at an orchard block level for compliance purposes, 80 attributes (independent variables) were created in three categories from the spray diary data: (1) individual insecticide applications applied during 2-week time windows, (2) groups of insecticide applications within time periods prior to or after fruit set and (3) orchard management attributes. Five machine learning algorithms (Decision Tree, Naïve Bayes, Random Forest, AdaBoost, Support Vector Machine) and one statistical method (Logistic regression) (classifiers) were used to develop models to forecast insecticide application decisions for leafroller control, by predicting whether pest monitoring results were above or below a spray threshold. Models to forecast 2011 spraying decisions were trained on 2008 and 2009 data and tested on 2010 data. Forecasts were made for spray and no-spray decisions based upon pre-determined acceptable rates of precision (proportion of correct decisions in test results). Orchard blocks in which a forecast could not be made to a prescribed degree of precision were recommended to be monitored, which is the normal practice. Spray decisions could not be forecast to an acceptable degree of precision, but decisions not to spray were successfully forecast for 49% of the blocks to a precision of 98% (AdaBoost) and 70% of the blocks to a precision of 95% (Naïve Bayes). Models with as few as four attributes gave useful forecasts, and orchard management attributes were the most important determinants of model forecasting accuracy. The potential for this methodology to assist with pest spray forecasting using customised data sets is discussed.

Mapping Techniques of Spatiotemporal Relationships for a Centric Road Network Model

The aim of the study is to determine the effectiveness of cartographic representation methods based on the proportion of correct decisions for each map, as well as users’ perceived effectiveness and their subjective evaluation of the graphical attractiveness of the mapping techniques used to convey information about spatio temporal accessibility. Several visualisations (both traditional and less conventional) are used to show distances between points in an hypothetical centric road network model. The relationships between the starting point and the destination points are considered in three ways: the Euclidean distance, the topographic distance measured along the roads, and the temporal distance. Fifty students were asked to estimate shortest, longest and matching distances and to pick out from the options the most effective and the most graphically attractive method of portrayal. The best result was achieved for two-dimensional mapping techniques with information about accessibility level labelled directly on the map. Students also performed better when seeking the shortest distances between central and destination point, and the best estimation results were achieved for the Euclidean distance.

Validity of Alternative Cut-Off Scores for the Back-Saver Sit and Reach Test

The purpose of the study was to determine if currently used FITNESSGRAM® cut-off scores for the Back Saver Sit and Reach Test had the best criterion-referenced validity evidence for 6–12 year old children. Secondary analyses of an existing data set focused on the passive straight leg raise and Back Saver Sit and Reach Test flexibility scores of 87 boys and 91 girls. The reliability and validity evidence was determined for Back Saver Sit and Reach Test cut-off scores of 7, 8, 9, and 10 inches for the boys and girls separately. Reliability evidence for passive straight leg raise and all Back Saver Sit and Reach Test cut-off scores was good to excellent. The most valid cut-off scores had the highest proportion of correct decisions associated with sensitivity ≥.67, and false-positive rate ≤.24. Based on moderate validity evidence, the best Back Saver Sit and Reach Test cut-off scores for 6–12 year olds are 8 and 9 inches for boys and girls, respectively. These findings should be cross-validated with other samples.

Binomial Sequential Sampling for Management of Aster Leafhopper (Homoptera: Cicadellidae) and Aster Yellows Phytoplasma in Carrot: Impact of Tally Threshold on the Accuracy of Treatment Decisions

We sampled aster leafhopper, Macrosteles quadrilineatus (Forbes), populations in 1996 and 1997 from central and southern Minnesota to develop a binomial sequential sampling plan for the leafhopper in carrot, Daucus carota (L.). Both conventional and organic farms were sampled with a sample unit consisting of 10 sweeps. Mean density ranged from 0.13 to 16.15 leafhoppers per 10 sweeps. Resampling simulation software was used with the Wald sequential probability ratio test to validate a binomial sampling plan from field data. Action thresholds were chosen based on aster yellows infectivity level (the proportion of individuals vectoring the phytoplasma) and varietal tolerance combinations. We used 2 infectivity levels (1 and 5%) and 3 varietal tolerance levels (resistant, intermediate, and susceptible) to represent 6 scenarios encountered in the field. Results from resampling validation analyses were used to select a final tally threshold (number of insects in a sample needed to consider the sample infested), based on the probability of making a decision to treat (i.e., operating characteristic). The operating characteristic was analyzed to determine which tally threshold provided the highest proportion of correct decisions. A tally threshold of 3 was selected as the overall optimum across infectivity level and varietal tolerance combinations. For the 6 possible infestation scenarios, 4–10 samples (10 sweeps each) were required to make a treat or no treat decision. Stop-line graphs are provided for field use for all 6 scenarios. Use of binomial sequential sampling plans should provide effective management of M. quadrilineatus in carrots and minimize sampling time and cost.

Group Dependence of Some Reliability Indices for Mastery Tests

Reliability indices for mastery tests depend not only on true-score variance but also on mean and cutoff scores. This dependence was examined in the case of three decision-theoretic indices: (1) the co efficient of agreement; (2) kappa; and (3) the pro portion of correct decisions. The binomial error model was assumed, with a two-parameter beta dis tribution for true scores. The reliability indices were computed at five values of the mean, four values of KR-21, and four cutoff scores. Results show that the dependence of kappa on mean and cutoff scores is opposite to that of the proportion of correct de cisions, which is linearly related to average thresh old loss. Moreover, kappa can be very small when most examinees are classified correctly. Thus, ob jections against the classical reliability coefficient apply even more strongly to kappa.