David Delgado-Gomez

Combining scales to assess suicide risk

OBJECTIVES A major interest in the assessment of suicide risk is to develop an accurate instrument, which could be easily adopted by clinicians. This article aims at identifying the most discriminative items from a collection of scales usually employed in the assessment of suicidal behavior. METHODS The answers to the Barrat Impulsiveness Scale, International Personality Disorder Evaluation Screening Questionnaire, Brown-Goodwin Lifetime History of Aggression, and Holmes & Rahe Social Readjustment Rating Scale provided by a group of 687 subjects (249 suicide attempters, 81 non-suicidal psychiatric inpatients, and 357 healthy controls) were used by the Lars-en algorithm to select the most discriminative items. RESULTS We achieved an average accuracy of 86.4%, a specificity of 89.6%, and a sensitivity of 80.8% in classifying suicide attempters using 27 out of the 154 items from the original scales. CONCLUSIONS The 27 items reported here should be considered a preliminary step in the development of a new scale evaluating suicidal risk in settings where time is scarce.

Suicide attempters classification: Toward predictive models of suicidal behavior

Suicide is a major public health issue with considerable human and economic cost. Previous attempts to delineate techniques capable of accurately predicting suicidal behavior proved unsuccessful. This paper aims at classifying suicide attempters (SA) as a first step toward the development of predictive models of suicidal behavior. A sample of 883 adults (347 SA and 536 non-SA) admitted to two university hospitals in Madrid, Spain, between 1999 and 2003 was used. Five multivariate techniques (linear regression, stepwise linear regression, decision trees, Lars-en and support vector machines) were compared with regard to their capacity to accurately classify SA. These techniques were applied to the Holmes-Rahe social readjustment rating scale and the international personal disorder examination screening questionnaire. Combining both scales, the Lars-en and stepwise linear regression techniques achieved 83.6% and 82.3% classification accuracy, respectively. In addition, these classification results were obtained using less than half of the available items. Multivariate techniques demonstrated to be useful in classifying SA using a combination of life events and personality criteria with reasonable accuracy, sensitivity and specificity.

Similarity-based Fisherfaces

In this article, a face recognition algorithm aimed at mimicking the human ability to differentiate people is proposed. For each individual, we first compute a projection line that maximizes his or her dissimilarity to all other people in the user database. Facial identity is thus encoded in the dissimilarity pattern composed by all the projection coefficients of an individual against all other enrolled user identities. Facial recognition is achieved by calculating the dissimilarity pattern of an unknown individual with that of each enrolled user. As the proposed algorithm is composed of different one-dimensional projection lines, it easily allows adding or removing users by simply adding or removing the corresponding projection lines in the system. Ideally, to minimize the influence of these additions/removals, the user group should be representative enough of the general population. Experiments on three widely used databases (XM2VTS, AR and Equinox) show consistently good results. The proposed algorithm achieves Equal Error Rate (EER) and Half-Total Error Rate (HTER) values in the ranges of 0.41-1.67% and 0.1-1.95%, respectively. Our approach yields results comparable to the top two winners in recent contests reported in the literature.

Can the Holmes-Rahe Social Readjustment Rating Scale (SRRS) Be Used as a Suicide Risk Scale? An Exploratory Study

The objective of this research was to examine whether the Holmes-Rahe Social Readjustment Rating Scale, a life event scale, can be used to identify suicide attempters. The Holmes-Rahe Social Readjustment Rating Scales ability to identify suicide attempters was tested in 1183 subjects (478 suicide attempters, 197 psychiatric inpatients, and 508 healthy controls) using the Fisher Linear Discriminant Analysis and traditional psychometric methods. The Fisher Linear Discriminant Analysis outperformed traditional psychometric approaches (area under the curve: 0.85 vs. 0.78; p < 0.05) and indicated that this scale may be used to identify suicide attempters. The life events that better characterized suicide attempters were change in frequency of arguments, marital separation, and personal injury. The Holmes-Rahe Social Readjustment Rating Scale may help identify suicide attempters.

User profiles of an electronic mental health tool for ecological momentary assessment: MEmind

Ecological momentary assessment (EMA) is gaining importance in psychiatry. This article assesses the characteristics of patients who used a new electronic EMA tool: the MEmind Wellness Tracker. Over one year, 13811 adult outpatients in our Psychiatry Department were asked to use MEmind. We collected information about socio‐demographic data, psychiatric diagnoses, illness severity, stressful life events and suicidal thoughts/behavior. We compared active users (N = 2838) and non‐active users (N = 10,973) of MEmind and performed a Random Forest analysis to assess which variables could predict its use. Univariate analyses revealed that MEmind‐users were younger (42.2 ± 13.5 years versus 48.5 ± 16.3 years; χ2 = 18.85; P < 0.001) and more frequently diagnosed with anxiety related disorders (57.9% versus 46.7%; χ2 = 105.92; P = 0.000) than non‐active users. They were more likely to report thoughts about death and suicide (up to 24% of active users expressed wish for death) and had experienced more stressful life events than non‐active users (57% versus 48.5%; χ2 = 64.65; P < 0.001). In the Random Forest analysis, 31 variables showed mean decrease accuracy values higher than zero with a 95% confidence interval (CI), including sex, age, suicidal thoughts, life threatening events and several diagnoses. In the light of these results, strategies to improve EMA and e‐Mental Health adherence are discussed.

Age of First Suicide Attempt in Men and Women: An Admixture Analysis

Objectives. To define different subgroups of suicide attempters according to age at onset of suicide attempts. Methods. Participants were 229 suicide attempters (147 females; 82 males) admitted to a general hospital in Madrid, Spain. We used admixture analysis to determine the best-fitting model for the age at onset of suicide attempts separated by sex. Results. The best fitted model for the age at onset of suicide attempts was a mixture of two gaussian distributions. Females showed an earlier age at onset of suicide attempts in both Gaussian distributions (mean ± S.D.) (26.98 ± 5.69 and 47.98 ± 14.13) than males (32.77 ± 8.11 and 61.31 ± 14.61). Early-onset female attempters were more likely to show borderline personality disorder than late-onset female attempters (OR = 11.11; 95% CI = 2.43–50.0). Conclusions. Age at onset of suicide attempts characterizes different subpopulations of suicide attempters.

Escala Abreviada de Personalidad y Acontecimientos Vitales para la detección de los intentos de suicidio

OBJECTIVE To develop a brief and reliable psychometric scale to identify individuals at risk for suicidal behaviour. METHOD DESIGN Case-control study. SAMPLE AND SETTING 182 individuals (61 suicide attempters, 57 psychiatric controls, and 64 psychiatrically healthy controls) aged 18 or older, admitted to the Emergency Department at Puerta de Hierro University Hospital in Madrid, Spain. MEASURES All participants completed a form including their socio-demographic and clinical characteristics, and the Personality and Life Events scale (27 items). To assess Axis I diagnoses, all psychiatric patients (including suicide attempters) were administered the Mini International Neuropsychiatric Interview. STATISTICAL ANALYSIS Descriptive statistics were computed for the socio-demographic factors. Additionally, χ(2) independence tests were applied to evaluate differences in socio-demographic and clinical variables, and the Personality and Life Events scale between groups. A stepwise linear regression with backward variable selection was conducted to build the Short Personality Life Event (S-PLE) scale. In order to evaluate the accuracy, a ROC analysis was conducted. The internal reliability was assessed using Cronbachs α, and the external reliability was evaluated using a test-retest procedure. RESULTS The S-PLE scale, composed of just 6 items, showed good performance in discriminating between medical controls, psychiatric controls and suicide attempters in an independent sample. For instance, the S-PLE scale discriminated between past suicide and past non-suicide attempters with sensitivity of 80% and specificity of 75%. The area under the ROC curve was 88%. A factor analysis extracted only one factor, revealing a single dimension of the S-PLE scale. Furthermore, the S-PLE scale provides values of internal and external reliability between poor (test-retest: 0.55) and acceptable (Cronbachs α: 0.65) ranges. Administration time is about one minute. CONCLUSIONS The S-PLE scale is a useful and accurate instrument for estimating the risk of suicidal behaviour in settings where the time is scarce.

Computerized Adaptive Test vs. decision trees: Development of a support decision system to identify suicidal behavior

BACKGROUND Several Computerized Adaptive Tests (CATs) have been proposed to facilitate assessments in mental health. These tests are built in a standard way, disregarding useful and usually available information not included in the assessment scales that could increase the precision and utility of CATs, such as the history of suicide attempts. METHODS Using the items of a previously developed scale for suicidal risk, we compared the performance of a standard CAT and a decision tree in a support decision system to identify suicidal behavior. We included the history of past suicide attempts as a class for the separation of patients in the decision tree. RESULTS The decision tree needed an average of four items to achieve a similar accuracy than a standard CAT with nine items. The accuracy of the decision tree, obtained after 25 cross-validations, was 81.4%. A shortened test adapted for the separation of suicidal and non-suicidal patients was developed. CONCLUSION CATs can be very useful tools for the assessment of suicidal risk. However, standard CATs do not use all the information that is available. A decision tree can improve the precision of the assessment since they are constructed using a priori information.

Restructuring bank networks after mergers and acquisitions

During restructuring processes, due to mergers and acquisitions, banks frequently face the problem of having redundant branches competing in the same market. In this work, we introduce a new Capacitated Branch Restructuring Model which extends the available literature in delocation models. It considers both closing down and long term operations? costs, and addresses the problem of resizing open branches in order to maintain a constant service level. We consider, as well, the presence of competitors and allow for ceding market share whenever the restructuring costs are prohibitively expensive.We test our model in a real life scenario, obtaining a reduction of about 40% of the network size, and annual savings over 45% in operation costs from the second year on. We finally perform a sensitivity analysis on critical parameters. This analysis shows that the final design of the network depends on certain strategic decisions concerning the redundancy of the branches, as well as their proximity to the demand nodes and to the competitor?s branches. At the same time, this design is quite robust to changes in the parameters associated with the adjustments on service capacity and with the market reaction.

An Experimental Evaluation of Three Classifiers for Use in Self-Updating Face Recognition Systems

Previous studies have shown that the accuracy of Face Recognition Systems (FRSs) decreases with the time elapsed between enrollment and testing. The main reason for the decrease is the changes in appearance of the user due to factors such as ageing, beard growth, sun-tan etc. Self-update procedure, where the system learns the biometric characteristics of the user every time he/she interacts with it, can be used to automatically update the system. However, a commonly acknowledged problem is the corruption of biometric traits due to misclassification. In this article, we test FRS, based on three classification algorithms, on two challenging databases, GEFA and YT, with 14 279 and 31 951 images, respectively. Our results suggest that complex, state-of-the-art classifiers that make use of user-specific models, need not be the best choice for use in self updating systems. In other words, tolerance to corrupted training data decreases as the complexity of the classifier increases.