Using data mining for prediction of hospital length of stay: an application of the CRISP-DM Methodology

Hospitals are nowadays collecting vast amounts of data related with patient records. All this data hold valuable knowledge that can be used to improve hospital decision making. Data mining techniques aim precisely at the extraction of useful knowledge from raw data. This work describes an implementation of a medical data mining project approach based on the CRISP-DM methodology. Recent real-world data, from 2000 to 2013, were collected from a Portuguese hospital and related with inpatient hospitalization. The goal was to predict generic hospital Length Of Stay based on indicators that are commonly available at the hospitalization process (e.g., gender, age, episode type, medical specialty). At the data preparation stage, the data were cleaned and variables were selected and transformed, leading to 14 inputs. Next, at the modeling stage, a regression approach was adopted, where six learning methods were compared: Average Prediction, Multiple Regression, Decision Tree, Artificial Neural Network ensemble, Support Vector Machine and Random Forest. The best learning model was obtained by the Random Forest method, which presents a high quality coefficient of determination value (0.81). This model was then opened by using a sensitivity analysis procedure that revealed three influential input attributes: the hospital episode type, the physical service where the patient is hospitalized and the associated medical specialty. Such extracted knowledge confirmed that the obtained predictive model is credible and with potential value for supporting decisions of hospital managers.

Risk Factors for Cardiovascular Disease, Metabolic Syndrome and Sleepiness in Truck Drivers

Abstract Background: Truck driver sleepiness is a primary cause of vehicle accidents. Several causes are associated with sleepiness in truck drivers. Obesity and metabolic syndrome (MetS) are associated with sleep disorders and with primary risk factors for cardiovascular diseases (CVD). We analyzed the relationship between these conditions and prevalence of sleepiness in truck drivers. Methods: We analyzed the major risk factors for CVD, anthropometric data and sleep disorders in 2228 male truck drivers from 148 road stops made by the Federal Highway Police from 2006 to 2011. Alcohol consumption, illicit drugs and overtime working hours were also analyzed. Sleepiness was assessed using the Epworth Sleepiness Scale. Results: Mean age was 43.1 ± 10.8 years. From 2006 to 2011, an increase in neck (p = 0.011) and abdominal circumference (p < 0.001), total cholesterol (p < 0.001), triglyceride plasma levels (p = 0.014), and sleepiness was observed (p < 0.001). In addition, a reduction in hypertension (39.6% to 25.9%, p < 0.001), alcohol consumption (32% to 23%, p = 0.033) and overtime hours (52.2% to 42.8%, p < 0.001) was found. Linear regression analysis showed that sleepiness correlated closely with body mass index (β = 0.19, Raj2 = 0.659, p = 0.031), abdominal circumference (β = 0.24, Raj2 = 0.826, p = 0.021), hypertension (β = -0.62, Raj2 = 0.901, p = 0.002), and triglycerides (β = 0.34, Raj2 = 0.936, p = 0.022). Linear multiple regression indicated that hypertension (p = 0.008) and abdominal circumference (p = 0.025) are independent variables for sleepiness. Conclusions: Increased prevalence of sleepiness was associated with major components of the MetS.

Factors affecting phosphate fixation in soils of the state of São Paulo

A laboratory experiment was carried out to study the effects of chemical and physical characteristics of the soil on the phosphate fixing capacity. One hundred samples collected from various localities were at first characterized chemically and their particlesize distribution determined. They were then tested as to their phosphate fixing capacities. The results obtained were statistically analysed by means of both simple linear and multiple correlation. The following conclusions could be drawn: 1. simple linear regression analysis indicated that % C, exchangeable Al+3, CEC, % clay, pH and % sand were the soil characteristics which significantly affected phosphate fixing capacity of São Paulo State soils; 2. multiple linear regression analysis indicated that % C, exchangeable Mg(+2)9 exchangeable- Al+3 and % clay were the soil characteristics which significantly affected the phosphate fixing capacity of São Paulo State soils; 3. the phosphate phonomena fixing as they occur in the soils of the São Paulo State can be best described by the following equation: Y = -2,266 - 3,484 + 3,514 + 5,559 + 1,005 %C Mg+2 Al+3 % clay exchangeable exchangeable 4. phosphate fixation in the soil is affected by the combined effects of both soil chemical and physical characteristics.

Morphometric analysis of Capitella capitata (Polychaeta, Capitellidae)

Accurate size measurements are fundamental in characterizing the population structure and secondary production of a species. The purpose of this study was to determine the best morphometric parameter to estimate the size of individuals of Capitella capitata (Fabricius, 1780). The morphometric analysis was applied to individuals collected in the intertidal zones of two beaches on the northern coast of the state of São Paulo, Brazil: São Francisco and Araçá. The following measurements were taken: the width and length (height) of the 4th, 5th and 7th setigers, and the length of the thoracic region (first nine setigers). The area and volume of these setigers were calculated and a linear regression analysis was applied to the data. The data were log-transformed to fit the allometric equation y = ax b into a straight line (log y = log a + b * log x). The measurements which best correlated with the thoracic length in individuals from both beaches were the length of setiger 5 (r² = 0.722; p<0.05 in São Francisco and r² = 0.795; p<0.05 in Araçá) and the area of setiger 7 (r² = 0.705; p<0.05 in São Francisco and r² = 0.634; p<0.05 in Araçá). According to these analyses, the length of setiger 5 and/or the area of setiger 7 are the best parameters to evaluate the growth of individuals of C. capitata.

Predicting hematologic toxicity in patients undergoing radioimmunotherapy with 90Y-ibritumomab tiuxetan or 131I-tositumomab.

This study aimed at identifying clinical factors for predicting hematologic toxicity after radioimmunotherapy with (90)Y-ibritumomab tiuxetan or (131)I-tositumomab in clinical practice. Hematologic data were available from 14 non-Hodgkin lymphoma patients treated with (90)Y-ibritumomab tiuxetan and 18 who received (131)I-tositumomab. The percentage baseline at nadir and 4 wk post nadir and the time to nadir were selected as the toxicity indicators for both platelets and neutrophils. Multiple linear regression analysis was performed to identify significant predictors (P < 0.05) of each indicator. For both platelets and neutrophils, pooled and separate analyses of (90)Y-ibritumomab tiuxetan and (131)I-tositumomab data yielded the time elapsed since the last chemotherapy as the only significant predictor of the percentage baseline at nadir. The extent of bone marrow involvement was not a significant factor in this study, possibly because of the short time elapsed since the last chemotherapy of the 7 patients with bone marrow involvement. Because both treatments were designed to deliver a comparable bone marrow dose, this factor also was not significant. None of the 14 factors considered was predictive of the time to nadir. The R(2) value for the model predicting percentage baseline at nadir was 0.60 for platelets and 0.40 for neutrophils. This model predicted the platelet and neutrophil toxicity grade to within ±1 for 28 and 30 of the 32 patients, respectively. For the 7 patients predicted with grade I thrombocytopenia, 6 of whom had actual grade I-II, dosing might be increased to improve treatment efficacy. The elapsed time since the last chemotherapy can be used to predict hematologic toxicity and customize the current dosing method in radioimmunotherapy.

Challenging recommended oral and intravenous voriconazole doses for improved efficacy and safety: population pharmacokinetics-based analysis of adult patients with invasive fungal infections.

BACKGROUND: Recommended oral voriconazole (VRC) doses are lower than intravenous doses. Because plasma concentrations impact efficacy and safety of therapy, optimizing individual drug exposure may improve these outcomes. METHODS: A population pharmacokinetic analysis (NONMEM) was performed on 505 plasma concentration measurements involving 55 patients with invasive mycoses who received recommended VRC doses. RESULTS: A 1-compartment model with first-order absorption and elimination best fitted the data. VRC clearance was 5.2 L/h, the volume of distribution was 92 L, the absorption rate constant was 1.1 hour(-1), and oral bioavailability was 0.63. Severe cholestasis decreased VRC elimination by 52%. A large interpatient variability was observed on clearance (coefficient of variation [CV], 40%) and bioavailability (CV 84%), and an interoccasion variability was observed on bioavailability (CV, 93%). Lack of response to therapy occurred in 12 of 55 patients (22%), and grade 3 neurotoxicity occurred in 5 of 55 patients (9%). A logistic multivariate regression analysis revealed an independent association between VRC trough concentrations and probability of response or neurotoxicity by identifying a therapeutic range of 1.5 mg/L (>85% probability of response) to 4.5 mg/L (<15% probability of neurotoxicity). Population-based simulations with the recommended 200 mg oral or 300 mg intravenous twice-daily regimens predicted probabilities of 49% and 87%, respectively, for achievement of 1.5 mg/L and of 8% and 37%, respectively, for achievement of 4.5 mg/L. With 300-400 mg twice-daily oral doses and 200-300 mg twice-daily intravenous doses, the predicted probabilities of achieving the lower target concentration were 68%-78% for the oral regimen and 70%-87% for the intravenous regimen, and the predicted probabilities of achieving the upper target concentration were 19%-29% for the oral regimen and 18%-37% for the intravenous regimen. CONCLUSIONS: Higher oral than intravenous VRC doses, followed by individualized adjustments based on measured plasma concentrations, improve achievement of the therapeutic target that maximizes the probability of therapeutic response and minimizes the probability of neurotoxicity. These findings challenge dose recommendations for VRC.

Determinació de la deposició atmosfèrica de metalls pesants a les Muntanyes de Prades (Tarragona) a partir dels bioindicadors Xanthoria parietina i Hypnum cupressiforme

En aquest estudi s’ha determinat la deposició atmosfèrica de metalls pesants a partir dels bioindicadors Xanthoria parietina (liquen) i Hypnum cupressiforme (molsa), en 16 punts de mostreig repartits per la zona on s’ubicarà el futur parc natural de les Muntanyes de Prades (Tarragona). A partir de la comparació de les concentracions obtingudes en aquest treball amb les d’altres estudis de plantejament equiparable, es va comprovar que la majoria de valors de concentració dels metalls eren similars als d’altres zones geogràfiques estudiades, a excepció del crom i el níquel, que presentaven valors clarament més elevats. A partir de proves estadístiques (ANOVA, regressions múltiples) es va determinar que existien diferències significatives entre les concentracions de metalls dels diferents punts de mostreig. També es van detectar correlacions significatives entre les concentracions dels punts de mostreig i la distància d’aquests respecte els focus potencials de contaminació que s’havien considerat prèviament. Amb les dades obtingudes a partir un petit anàlisi de la riquesa liquènica dels punts de mostreig, es va disposar d’informació extra per poder avaluar les fonts i el patró de distribució d’aquests contaminants.

Okun’s law – A meta analysis

This paper seeks to identify whether there is a representative empirical Okun’s Law coefficient (OLC) and to measure its size. We carry out a meta regression analysis on a sample of 269 estimates of the OLC to uncover reasons for differences in empirical results and to estimate the ‘true’ OLC. On statistical (and other) grounds, we find it appropriate to investigate two separate subsamples, using respectively (some measure of) unemployment or output as dependent variable. Our results can be summarized as follows. First, there is evidence of type II publication bias in both sub-samples, but a type I bias is present only among the papers using some measure of unemployment as the dependent variable. Second, after correction for publication bias, authentic and statistically significant OLC effects are present in both sub-samples. Third, bias-corrected estimated true OLCs are significantly lower (in absolute value) with models using some measure of unemployment as the dependent variable. Using a bivariate MRA approach, the estimated true effects are -0.25 for the unemployment sub-sample and -0.61 for the output-sub sample; with a multivariate MRA methodology, the estimated true effects are -0.40 and -1.02 for the unemployment and the output-sub samples respectively.

A general model to predict individual exposure to solar UV by using ambient irradiance data

Excessive exposure to solar ultraviolet (UV) is the main cause of skin cancer. Specific prevention should be further developed to target overexposed or highly vulnerable populations. A better characterisation of anatomical UV exposure patterns is however needed for specific prevention. To develop a regression model for predicting the UV exposure ratio (ER, ratio between the anatomical dose and the corresponding ground level dose) for each body site without requiring individual measurements. A 3D numeric model (SimUVEx) was used to compute ER for various body sites and postures. A multiple fractional polynomial regression analysis was performed to identify predictors of ER. The regression model used simulation data and its performance was tested on an independent data set. Two input variables were sufficient to explain ER: the cosine of the maximal daily solar zenith angle and the fraction of the sky visible from the body site. The regression model was in good agreement with the simulated data ER (R(2)=0.988). Relative errors up to +20% and -10% were found in daily doses predictions, whereas an average relative error of only 2.4% (-0.03% to 5.4%) was found in yearly dose predictions. The regression model predicts accurately ER and UV doses on the basis of readily available data such as global UV erythemal irradiance measured at ground surface stations or inferred from satellite information. It renders the development of exposure data on a wide temporal and geographical scale possible and opens broad perspectives for epidemiological studies and skin cancer prevention.

Mixture of bivariate Poisson regression models with an application to insurance

In a recent paper Bermúdez [2009] used bivariate Poisson regression models for ratemaking in car insurance, and included zero-inflated models to account for the excess of zeros and the overdispersion in the data set. In the present paper, we revisit this model in order to consider alternatives. We propose a 2-finite mixture of bivariate Poisson regression models to demonstrate that the overdispersion in the data requires more structure if it is to be taken into account, and that a simple zero-inflated bivariate Poisson model does not suffice. At the same time, we show that a finite mixture of bivariate Poisson regression models embraces zero-inflated bivariate Poisson regression models as a special case. Additionally, we describe a model in which the mixing proportions are dependent on covariates when modelling the way in which each individual belongs to a separate cluster. Finally, an EM algorithm is provided in order to ensure the models’ ease-of-fit. These models are applied to the same automobile insurance claims data set as used in Bermúdez [2009] and it is shown that the modelling of the data set can be improved considerably.

A logistic regression approach to estimating customer profit loss due to lapses in insurance

This article focuses on business risk management in the insurance industry. A methodology for estimating the profit loss caused by each customer in the portfolio due to policy cancellation is proposed. Using data from a European insurance company, customer behaviour over time is analyzed in order to estimate the probability of policy cancelation and the resulting potential profit loss due to cancellation. Customers may have up to two different lines of business contracts: motor insurance and other diverse insurance (such as, home contents, life or accident insurance). Implications for understanding customer cancellation behaviour as the core of business risk management are outlined.

Ten years of lung transplantation in Switzerland: results of the Swiss Lung Transplant Registry.

Lung transplantation has evolved from an experimental procedure to a viable therapeutic option in many countries. In Switzerland, the first lung transplant was performed in November 1992, more than ten years after the first successful procedure world-wide. Thenceforward, a prospective national lung transplant registry was established, principally to enable quality control. The data of all patients transplanted in the two Swiss Lung Transplant centres Zurich University Hospital and Centre de Romandie (Geneva-Lausanne) were analysed. In 10 years 242 lung transplants have been performed. Underlying lung diseases were cystic fibrosis including bronchiectasis (32%), emphysema (32%), parenchymal disorders (19%), pulmonary hypertension (11%) and lymphangioleiomyomatosis (3%). There were only 3% redo procedures. The 1, 5 and 9 year survival rates were 77% (95% CI 72-82), 64% (95% CI 57-71) and 56% (95% CI 45-67), respectively. The 5 year survival rate of patients transplanted since 1998 was 72% (95% CI 64-80). Multivariate Cox regression analysis revealed that survival was significantly better in this group compared to those transplanted before 1998 (HR 0.44, 0.26-0.75). Patients aged 60 years and older (HR 5.67, 95% CI 2.50-12.89) and those with pulmonary hypertension (HR 2.01, 95% CI 1.10-3.65) had a significantly worse prognosis The most frequent causes of death were infections (29%), bronchiolitis obliterans syndrome (25%) and multiple organ failure (14%). The 10-year Swiss experience of lung transplantation compares favourably with the international data. The best results are obtained in cystic fibrosis, pulmonary emphysema and parenchymal disorders.

Clinical presentation and survival of smear-positive pulmonary tuberculosis patients of a University General Hospital in a developing country

From January 1995 to August 1997 we evaluated prospectively the clinical presentation, laboratory findings and short-term survival of smear-positive pulmonary tuberculosis (TB) patients who sought care at our hospital. After providing informed, written consent, the patients were interviewed and laboratory tests were performed. Information about survivorship and death was collected through September 1998. Eighty-six smear-positive pulmonary TB patients were enrolled; 26.7% were HIV-seropositive. Seventeen HIV-seronegative pulmonary TB patients (19.8%) presented chronic diseases in addition to TB. In the multiple logistic regression analysis a CD4+ cell count <= 200 cell/mm³ was independently associated with HIV seropositivity. In the Cox regression model, fitted to all patients, HIV seropositivity and age > or = 50 years were independently associated with decreased survival. Among HIV-seronegative persons, the presence of an additional disease increased the risk of death of almost six-fold. Use of antiretroviral drugs was associated with a lower risk of death among HIV-seropositive smear-positive pulmonary TB patients (RH = 0.32, 95% CI 0.10-0.92). In our study smear-positive pulmonary TB patients had a low short-term survival rate that was strongly associated with HIV infection, age and co-morbidities. Therapy with antiretroviral drugs reduced the short-term risk of death among HIV-seropositive patients after TB diagnosis.

Prevalence of Toxocara infection in schoolchildren from the Butantã region, São Paulo, Brazil

Visceral larva migrans syndrome by Toxocara affects mainly children between 2 and 5 years of age, it is generally asymptomatic, and the seroprevalence varies from 3 to 86% in different countries. A total of 399 schoolchildren from 14 public schools of the Butantã region, São Paulo city, Brazil, were evaluated by Toxocara serology (enzyme-linked immunosorbent assay). Epidemiological data to the Toxocara infection obtained from a protocol were submitted to multiple logistic regression analysis for a risk profile definition. Blood was collected on filter paper by finger puncture, with all samples tested in duplicate. Considering titers > 1/160 as positive, the seroprevalence obtained was 38.8%. Among infected children, the mean age was 9.4 years, with a similar distribution between genders. A significant association was observed with the presence of onychophagia, residence with a dirty backyard, living in a slum, previous wheezing episodes, school attended, and family income (p < 0.05). All data, except "living in a slum", were considered to be determinant of a risk profile for the acquisition of Toxocara infection. A monthly income > 5 minimum salaries represented a protective factor, although of low relevance. Toxocara eggs were found in at least one of the soil samples obtained from five schools, with high prevalence of Toxocara infections, indicating the frequent soil contamination by this agent.

Die Säuglingssterblichkeit in der Schweiz: multivariate Betrachtung. [Perinatal mortality in Switzerland: a multivariate analysis]

A regression analysis using a linked file of all Swiss births und perinatal deaths 1979-1981 showed a significant relation between birthweight and canton. Sex of infant and multiplicity of birth were significant, too. For live births, marital and socio-economic status of mother and father relate to birthweight. Logistic regressions brought out relationships between the risk of stillbirth and occupation of father, nationality and marital status of mother, apart from birthweight. For live births, only sex and (weakly) marital status and rank of the child were influencial after correction for birthweight.