Predicting Sexual Recidivism

Sexual Abuse ◽  
2019 ◽  
Vol 32 (4) ◽  
pp. 375-399 ◽  
Author(s):  
Turgut Ozkan ◽  
Stephen J. Clipper ◽  
Alex R. Piquero ◽  
Michael Baglivio ◽  
Kevin Wolff
Keyword(s):  
Risk Assessment ◽  
Predictive Accuracy ◽  
Area Under The Curve ◽  
Roc Curves ◽  
Sexual Recidivism ◽  
Florida Department ◽  
Forest Models ◽  
Random Forest Models ◽  
Sex Offense ◽  
Official Records

The current study focuses on adolescents with sex offense histories and examines sexual reoffending patterns within 2 years of a prior sex offense. We employed inductive statistical models using archival official records maintained by the Florida Department of Juvenile Justice (FDJJ), which provides social, offense, placement, and risk assessment history data for all youth referred for delinquent behavior. The predictive accuracy of the random forest models is tested using receiver operator characteristic (ROC) curves, the area under the curve (AUC), and precision/recall plots. The strongest predictor of sexual recidivism was the number of prior felony and misdemeanor sex offenses. The AUC values range between 0.71 and 0.65, suggesting modest predictive accuracy of the models presented. These results parallel the existing literature on sexual recidivism and highlight the challenges associated with predicting sex offense recidivism. Furthermore, results inform risk assessment literature by testing various factors recorded by an official institution.

scholarly journals Testing the Static-99R as a Global Screen for Risk of Sex Crime Recidivism in a Norwegian Routine Sample

Sexual Abuse ◽  
2020 ◽  
pp. 107906322095119
Author(s):  
Ingeborg Jenssen Sandbukt ◽  
Torbjørn Skardhamar ◽  
Ragnar Kristoffersen ◽  
Christine Friestad
Keyword(s):  
Risk Assessment ◽  
Logistic Regression ◽  
Random Forest ◽  
Predictive Accuracy ◽  
Proportional Hazard ◽  
Random Forest Classification ◽  
Forest Classification ◽  
Sex Offense ◽  
Observation Period ◽  
Sex Crime

The Static-99R has been recommended for use as a first global screen for sorting out sex-convicted persons who are in need of further risk assessment. This study investigated the Static-99R’s predictive validity based on a nonselected Norwegian sample ( n = 858) of persons released from prison after having served a sex crime sentence. After a mean observation period of 2,183 days, 3.4% ( n = 29) had recidivated to a new sex offense. A higher number of recidivists were found among those with higher Static-99R total scores. The predictive contribution from each of the ten Static-99R risk items was investigated using standard logistic regression, proportional hazard regression, and random forest classification algorithm. The overall results indicate that the Static-99R is relevant as a risk screen in a Norwegian context, providing similar results concerning predictive accuracy as previous studies.

A Validation Study of the Child Pornography Offender Risk Tool (CPORT)

Sexual Abuse ◽  
2018 ◽  
Vol 31 (4) ◽  
pp. 456-476 ◽  
Author(s):  
Angela W. Eke ◽  
L. Maaike Helmus ◽  
Michael C. Seto
Keyword(s):  
Statistical Power ◽  
Predictive Accuracy ◽  
Sexual Offending ◽  
Child Pornography ◽  
Area Under The Curve ◽  
Risk Assessments ◽  
Sexual Offense ◽  
Sexual Recidivism ◽  
Development And Validation

The Child Pornography Offender Risk Tool (CPORT) is a seven-item structured tool to assess the likelihood of future sexual offending over a 5-year fixed follow-up. The current study examined 5-year fixed follow-up data (15% any new sexual offense, 9% any new child pornography offense) for a validation sample of 80 men convicted of child pornography offense(s). Although statistical power was low, results were comparable with the development sample: The CPORT had slightly lower predictive accuracy for sexual recidivism for the overall group (area under the curve [AUC] = .70 vs. .74), but these values were not significantly different. Combining the development and validation samples, the CPORT predicted any sexual recidivism (AUC = .72) and child pornography recidivism specifically (AUC = .74), with similar accuracies. CPORT was also significantly predictive of these outcomes for the child pornography offenders with no known contact offenses. Strengths and weaknesses of incorporating CPORT into applied risk assessments are discussed.

scholarly journals A predictive model to estimate survival of hospitalized COVID-19 patients from admission data

2020 ◽  
Author(s):  
Todd J. Levy ◽  
Safiya Richardson ◽  
Kevin Coppa ◽  
Douglas P. Barnaby ◽  
Thomas McGinn ◽  
...  
Keyword(s):  
Survival Probability ◽  
Predictive Accuracy ◽  
Area Under The Curve ◽  
Predictive Performance ◽  
Roc Curves ◽  
Severity Index ◽  
Primary Objective ◽  
Lasso Regression ◽  
Distribution Width ◽  
Serum Blood Urea Nitrogen

AbstractObjectiveOur primary objective was to use initial data available to clinicians to characterize and predict survival for hospitalized coronavirus disease 2019 (COVID-19) patients. While clinical characteristics and mortality risk factors of COVID-19 patients have been reported, a practical survival calculator based on data from a diverse group of U.S. patients has not yet been introduced. Such a tool would provide timely and valuable guidance in decision-making during this global pandemic.DesignWe extracted demographic, laboratory, clinical, and treatment data from electronic health records and used it to build and test the predictive accuracy of a survival probability calculator referred to as “the Northwell COVID-19 Survival (‘NOCOS’) calculator.”Setting13 acute care facilities at Northwell Health served as the setting for this study.Participants5,233 hospitalized COVID-19–positive patients served as the participants for this study.Main outcome measuresThe NOCOS calculator was constructed using multivariate regression with L1 regularization (LASSO) to predict survival during hospitalization. Model predictive performance was measured using Receiver Operating Characteristic (ROC) curves and the Area Under the Curve (AUC) of the calculators.ResultsPatient age, serum blood urea nitrogen, Emergency Severity Index, red cell distribution width, absolute neutrophil count, serum bicarbonate, and glucose were identified as the optimal predictors of survival by multivariate LASSO regression. The predictive performance of the NOCOS calculator had an AUC of 0.832, reaching 0.91 when updated for each patient daily, with stability assessed and maintained for 14 consecutive days. This outperformed other established models, including the Sequential Organ Failure Assessment (SOFA) score (0.732).ConclusionsWe present a practical estimate of survival probability that outperforms other general risk models. The seven early predictors of in-hospital survival can help clinicians identify patients with increased probabilities of survival and provide critical decision support as COVID-19 spreads across the U.S.Trial registrationN/A

The Predictive Validity of the Static-99, Static-99R, and Static-2002/R: Which One to Use?

Sexual Abuse ◽  
2017 ◽  
Vol 30 (8) ◽  
pp. 887-907 ◽  
Author(s):  
Sophie G. Reeves ◽  
James R. P. Ogloff ◽  
Melanie Simmons
Keyword(s):  
Decision Making ◽  
Predictive Validity ◽  
Sexual Offenders ◽  
Predictive Accuracy ◽  
Risk Measures ◽  
Sexual Offending ◽  
Area Under The Curve ◽  
Sexual Recidivism ◽  
Sexual Offenses ◽  
Risk Categories

The use of Static tools (Static-99, Static-99R, Static-2002, and Static-2002R) in risk decision making involving sexual offenders is widespread internationally. This study compared the predictive accuracy and incremental validity of four Static risk measures in a sample of 621 Australian sexual offenders. Results indicated that approximately 45% of the sample recidivated (with 18.8% committing sexual offenses). All of the Static measures investigated yielded moderate predictive validity for sexual recidivism, which was comparable with other Australian and overseas studies. Area under the curve (AUC) values for the four measures across the 5-, 10-, and 15-year intervals ranged from .67 to .69. All of the Static measures discriminated quite well between low-risk and high-risk sexual offenders but less well for the moderate risk categories. When pitted together, none of the tools accounted for additional variance in sexual recidivism, above and beyond what the other measures accounted for. The overall results provide support for the use of Static measures as a component of risk assessment and decision making with Australian sexual offending populations. The limitations of this study and recommendations for further research are also discussed.

Predicting First-Time Sexual Offending Among Prisoners Without a Prior Sex Offense History

2012 ◽  
Vol 39 (11) ◽  
pp. 1436-1456 ◽  
Author(s):  
Grant Duwe
Keyword(s):  
Risk Assessment ◽  
Assessment Tool ◽  
Sexual Offending ◽  
Area Under The Curve ◽  
Bootstrap Resampling ◽  
Criminal Offending ◽  
Male Offenders ◽  
Sex Offense ◽  
First Time ◽  
Selection Of

In an effort to reduce first-time sexual offending, this study focuses on the development of a risk assessment tool, the Minnesota Sexual Criminal Offending Risk Estimate (MnSCORE), designed for prisoners without any prior sexual offending history. Logistic regression modeling was used to develop the MnSCORE on a sample of 9,064 male offenders released from Minnesota prisons between 2003 and 2006. Bootstrap resampling was used to not only refine the selection of predictors but also internally validate the model. With an optimism-corrected area under the curve (AUC) of 0.763, the results showed the MnSCORE has moderately high predictive discrimination. Because the risk of first-time sexual offending was significantly lower for offenders who completed prison-based chemical dependency (CD) treatment, it is anticipated the MnSCORE may best be used as a trailer assessment to help better prioritize prisoners for CD treatment—both in prison and in the community following release.

Comparing Predictive Validity of JSORRAT-II and MEGA♪ With Sexually Abusive Youth in Long-Term Residential Custody

2017 ◽  
Vol 62 (10) ◽  
pp. 2937-2953 ◽  
Author(s):  
Lucinda A. Lee Rasmussen
Keyword(s):  
Risk Assessment ◽  
Area Under The Curve ◽  
Assessment Tools ◽  
Sexual Recidivism ◽  
Residential Treatment Facility ◽  
Sexually Abusive ◽  
Risk Assessment Tools ◽  
Representative Samples

This 6-year prospective study is the first to compare two psychometrically sound risk assessment tools for sexually abusive youth: JSORRAT-II and MEGA♪. Cross-validated on representative samples of over 500 youth, these measures have cutoff scores, allowing for a more exact assessment of risk. Study sample consisted of 129 male adjudicated adolescents housed in a secured residential treatment facility for sexually abusive youth. Receiver operating characteristic (ROC) analysis showed that MEGA♪ Risk Scale was mildly predictive of sexual recidivism over a 6-year period (mean follow-up = 15.6 months)—area under the curve (AUC) = .67; 95% confidence interval [CI] = [0.52, 0.82]; p < .015. JSORRAT-II was not predictive (AUC = .57; 95% CI = [0.42, 0.72]; p < .297). The study contributes to scant literature on the most contemporary, statistically robust risk assessment tools for sexually abusive youth.

scholarly journals A Multimodality Machine Learning Approach to Differentiate Severe and Nonsevere COVID-19: Model Development and Validation

10.2196/23948 ◽  
2021 ◽  
Vol 23 (4) ◽  
pp. e23948
Author(s):  
Yuanfang Chen ◽  
Liu Ouyang ◽  
Forrest S Bao ◽  
Qian Li ◽  
Lei Han ◽  
...  
Keyword(s):  
Machine Learning ◽  
Random Forest ◽  
Laboratory Test ◽  
Predictive Accuracy ◽  
Learning Approach ◽  
Test Results ◽  
Laboratory Test Results ◽  
Machine Learning Approach ◽  
Forest Models ◽  
Random Forest Models

Background Effectively and efficiently diagnosing patients who have COVID-19 with the accurate clinical type of the disease is essential to achieve optimal outcomes for the patients as well as to reduce the risk of overloading the health care system. Currently, severe and nonsevere COVID-19 types are differentiated by only a few features, which do not comprehensively characterize the complicated pathological, physiological, and immunological responses to SARS-CoV-2 infection in the different disease types. In addition, these type-defining features may not be readily testable at the time of diagnosis. Objective In this study, we aimed to use a machine learning approach to understand COVID-19 more comprehensively, accurately differentiate severe and nonsevere COVID-19 clinical types based on multiple medical features, and provide reliable predictions of the clinical type of the disease. Methods For this study, we recruited 214 confirmed patients with nonsevere COVID-19 and 148 patients with severe COVID-19. The clinical characteristics (26 features) and laboratory test results (26 features) upon admission were acquired as two input modalities. Exploratory analyses demonstrated that these features differed substantially between two clinical types. Machine learning random forest models based on all the features in each modality as well as on the top 5 features in each modality combined were developed and validated to differentiate COVID-19 clinical types. Results Using clinical and laboratory results independently as input, the random forest models achieved >90% and >95% predictive accuracy, respectively. The importance scores of the input features were further evaluated, and the top 5 features from each modality were identified (age, hypertension, cardiovascular disease, gender, and diabetes for the clinical features modality, and dimerized plasmin fragment D, high sensitivity troponin I, absolute neutrophil count, interleukin 6, and lactate dehydrogenase for the laboratory testing modality, in descending order). Using these top 10 multimodal features as the only input instead of all 52 features combined, the random forest model was able to achieve 97% predictive accuracy. Conclusions Our findings shed light on how the human body reacts to SARS-CoV-2 infection as a unit and provide insights on effectively evaluating the disease severity of patients with COVID-19 based on more common medical features when gold standard features are not available. We suggest that clinical information can be used as an initial screening tool for self-evaluation and triage, while laboratory test results should be applied when accuracy is the priority.

Neutrophil-lymphocyte score: A novel prognostic scoring system that utilized the dynamic change of neutrophil, lymphocyte, and albumin and its comparison to other indices.

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. 3048-3048
Author(s):  
Songzhu Zhao ◽  
Mingjia Li ◽  
Daniel Spakowicz ◽  
Sandip H. Patel ◽  
Andrew Johns ◽  
...  
Keyword(s):  
Scoring System ◽  
Prognostic Value ◽  
Predictive Accuracy ◽  
Dynamic Change ◽  
Area Under The Curve ◽  
Roc Curves ◽  
Time Dependent ◽  
Advanced Lung Cancer ◽  
Lymphocyte Ratio ◽  
Lymphocyte To Monocyte Ratio

3048 Background: Indications for immune checkpoint inhibitor (ICI) in cancer care are expanding rapidly. There is increasing need for accurate decision tool to better guide treatment. We have constructed a new prognostic scoring system, neutrophil-lymphocyte score (NRS), based on the nonlinear dynamic change of neutrophil to lymphocyte ratio (NLR) in relation to survival over the first cycle of ICI treatment. We compared this novel system to existing indices such as NLR, lymphocyte to monocyte ratio (LMR), platelet to lymphocyte ratio (PLR), Advanced Lung Cancer Inflammation Index (ALI), and Systemic Immune-inflammation Index (SII). Methods: This is a retrospective analysis of 837 patients at Ohio State University from 2011-18. Neutrophil (ANC), lymphocyte (ALC), platelet (plt), monocyte (AMC), albumin (alb), and body mass index (BMI) were collected at baseline. Repeat labs were collected at cycle 2. NLR = ANC/ALC, ALI = BMI x alb / NLR, LMR = ALC/AMC, SII = platelet x NLR, PLR = plt/ALC. NLR Ratio = baseline NLR / repeat NLR. Based on the association between NLR and the overall survival, we assigned 1 point (p) for baseline NLR < 0.7, 6p for 0.7 to < 2, 5p for 2 to < 3, 4 p for 3 to < 4, 3 for 4 to 5, 2p for 5 to < 9, and 1p for ≥9. We also assigned 1p for NLR ratio < 0.6, 2p for 0.6 to < 0.8, 3p for 0.8 to < 1.2, 5p for 1.25 to < 1.4, 3p for 1.4 to < 1.6, and 2p for ≥1.6. NLS = sum of these 2 scores . NLS_A = NLS*alb. Time-dependent receiver operator characteristic (ROC) curves with integrated time-dependent area under the curve (TD AUC) values were used to evaluate the predictive accuracy of each index for survival. Results: For baseline and repeat values, all indices were statistically significant (P < 0.001) in predicting survival. Baseline integrated TD AUC were: ALI 0.704, NLR 0.692, SII 0.663, LMR 0.645, and PLR 0.612. All of the repeat indices at cycle 2 had higher prognostic value than their baseline counterparts. Integrated TD AUC for indices at cycle 2 were: ALI 0.740 (with baseline BMI), NLR 0.729, SII 0.694, LMR 0.671, and PLR 0.652. NLS_A was a composite score based on the dynamic change of NLR from cycle 1 to 2 and the treatment alb with integrated TD-AUC at 0.754. Conclusions: Indices constructed from ANC, ALC, AMC, Plt, alb, and BMI can be obtained inexpensively and provide great prognostic value for pts on ICI. We have constructed a novel scoring system (NLS_A) and demonstrated its improvement over the current prognostic indices. Studies with a larger cohort are needed to further improve and validate this system.

scholarly journals Treatment-Specific Composition of the Gut Microbiota Is Associated With Disease Remission in a Pediatric Crohn’s Disease Cohort

2019 ◽  
Vol 25 (12) ◽  
pp. 1927-1938 ◽  
Author(s):  
Daniel Sprockett ◽  
Natalie Fischer ◽  
Rotem Sigall Boneh ◽  
Dan Turner ◽  
Jarek Kierkus ◽  
...  
Keyword(s):  
Crohn’S Disease ◽  
Random Forest ◽  
Crohn's Disease ◽  
Gut Microbiota ◽  
Area Under The Curve ◽  
Disease Remission ◽  
Beneficial Effects ◽  
Forest Models ◽  
Random Forest Models ◽  
The Impact

Abstract Background The beneficial effects of antibiotics in Crohn’s disease (CD) depend in part on the gut microbiota but are inadequately understood. We investigated the impact of metronidazole (MET) and metronidazole plus azithromycin (MET+AZ) on the microbiota in pediatric CD and the use of microbiota features as classifiers or predictors of disease remission. Methods 16S rRNA-based microbiota profiling was performed on stool samples from 67 patients in a multinational, randomized, controlled, longitudinal, 12-week trial of MET vs MET+AZ in children with mild to moderate CD. Profiles were analyzed together with disease activity, and then used to construct random forest models to classify remission or predict treatment response. Results Both MET and MET+AZ significantly decreased diversity of the microbiota and caused large treatment-specific shifts in microbiota structure at week 4. Disease remission was associated with a treatment-specific microbiota configuration. Random forest models constructed from microbiota profiles before and during antibiotic treatment with metronidazole accurately classified disease remission in this treatment group (area under the curve [AUC], 0.879; 95% confidence interval, 0.683–0.9877; sensitivity, 0.7778; specificity, 1.000; P &lt; 0.001). A random forest model trained on pre-antibiotic microbiota profiles predicted disease remission at week 4 with modest accuracy (AUC, 0.8; P = 0.24). Conclusions MET and MET+AZ antibiotic regimens in pediatric CD lead to distinct gut microbiota structures at remission. It may be possible to classify and predict remission based in part on microbiota profiles, but larger cohorts will be needed to realize this goal.

The Ontario Domestic Assault Risk Assessment: Predicting Violence Among Men With a Police Record of Intimate Partner Violence in the United States

2021 ◽  
pp. 009385482110358
Author(s):  
Dana L. Radatz ◽  
N. Zoe Hilton
Keyword(s):  
Risk Assessment ◽  
Intimate Partner Violence ◽  
Partner Violence ◽  
Assessment Tool ◽  
Predictive Accuracy ◽  
Area Under The Curve ◽  
The United States ◽  
Intimate Partner ◽  
Domestic Assault ◽  
Violent Recidivism

The Ontario Domestic Assault Risk Assessment (ODARA) is an actuarial risk assessment tool for intimate partner violence (IPV) recidivism. Despite its international use, there is no published validation of the ODARA’s predictive accuracy in a U.S. sample. We studied 356 men in New York police records of IPV against a female partner to examine the ODARA’s predictive accuracy for IPV recidivism (base rate 35%), non-IPV violent recidivism (against a nonpartner; 16%), any violent recidivism (49%), and nonviolent recidivism (50%), in a fixed 2-year follow-up. Using 11 scorable ODARA items, area under the curve values were significant and ranged from .590 to .630, indicating small to medium effects. Expected/Observed indices revealed poor calibration with 2-year IPV recidivism rates in ODARA construction and cross-validation samples. Findings support the generalization of the ODARA’s predictive accuracy in different populations and outcomes, but a need for new norm development for higher risk populations.

Sign in / Sign up

Export Citation Format

Share Document