Predicting Overall Survival in Patients with Metastatic Rectal Cancer: a Machine Learning Approach

BACKGROUND

A significant proportion of patients with rectal cancer will present with synchronous metastasis at the time of diagnosis. Overall survival (OS) for these patients are highly variable and previous attempts to build predictive models often have low predictive power, with concordance indexes (c-index) less than 0.70.

METHODS

Using the National Cancer Database (2010-2014), we identified patients with synchronous metastatic rectal cancer. The data was split into a training dataset (diagnosis years 2010-2012), which was used to build the machine learning model, and a testing dataset (diagnosis years 2013-2014), which was used to externally validate the model. A nomogram predicting 3-year OS was created using Cox proportional hazard regression with lasso penalization. Predictors were selected based on clinical significance and availability in NCDB. Performance of the machine learning model was assessed by c-index.

RESULTS

A total of 4098 and 3107 patients were used to construct and validate the nomogram, respectively. Internally validated c-indexes at 1, 2, and 3 years were 0.816 (95% CI 0.813-0.818), 0.789 (95% CI 0.786-0.790), and 0.778 (95% CI 0.775-0.780), respectively. External validated c-indexes at 1, 2, and 3 years were 0.811, 0.779, and 0.778, respectively.

CONCLUSIONS

There is wide variability in the OS for patients with metastatic rectal cancer, making accurate predictions difficult. However, using machine learning techniques, more accurate models can be built. This will aid patients and clinicians in setting expectations and making clinical decisions in this group of challenging patients.

Evaluation of drinking patterns and their impact on alcohol-related aggression: a national survey of adolescent behaviours

Background

Although there have been a wide range of epidemiological studies examining the impact of patterns of alcohol consumption among adolescents, there remains considerable variability in both defining these patterns and the ability to comprehensively evaluate their relationship to behavioural patterns. This study explores a new procedure for defining and evaluating drinking patterns and integrating well-established indicators. The composite measure is then used to estimate the impact of these patterns on alcohol-related aggressive behaviour among Italian adolescents.

Methods

Data were collected as part of the 2011 European School Survey Project on Alcohol and other Drugs (ESPAD). A national sample of 14,199 students aged 15–19 years was collected using an anonymous, self-administered questionnaire completed in a classroom setting. Drinking patterns were established using principal component analysis. Alcohol-related aggression was analysed as to its relationship to patterns of drinking, behaviour of friends towards alcohol use, substance use/abuse, school performance, family relationships and leisure activities.

Results

Several specific drinking patterns were identified: “Drinking to Excess” (DE), “Drinking with Intoxication” (DI) and “Drinking but Not to Excess” (DNE). A higher percentage of males were involved in alcohol-related aggression compared with females. In males, the DE and DI patterns significantly increased the likelihood of alcohol-related aggression, whereas the DNE pattern was negatively associated. Similar results were found in females, although the DI pattern was not significantly associated with alcohol-related aggression. Overall, cigarette smoking, illegal drug use, truancy, limited parental monitoring, frequent evenings spent outside of the home and peer influence associated strongly with alcohol-related aggression.

Conclusions

Our findings suggest that drinking patterns, as uniquely monitored with an integrated metric, can: 1) explain drinking habits better than commonly used indicators of alcohol use and 2) provide a better understanding of behavioural risks such as alcohol-related aggression. Environmental background also appears to strongly associate with this type of aggressive behaviour.

Alcohol use and patterns of delinquent behaviour in male and female adolescents

AIMS

The overall aim was to study patterns of delinquent behaviour in relation to adolescent alcohol use. The more specific aims were to examine whether alcohol use varied between groups of adolescents with different patterns of delinquent behaviour, and to explore whether the association between delinquent behaviour patterns and alcohol use was similar for males and females.

METHODS

The participants were male (n = 406) and female (n = 532) adolescents in the eighth grade (age 14 years) in a medium-sized city of Sweden. We used information about self-rated alcohol use and different types of delinquent behaviour.

RESULTS

The results revealed that the occurrence of excessive alcohol use and drunkenness varied between groups of adolescents with different delinquency patterns, and that the associations between alcohol use and patterns of delinquent behaviours were relatively similar for males and females. Adolescents with patterns characterized by more serious non-violent delinquency or by violent delinquency reported the highest occurrence of alcohol use and frequency of drunkenness. Adolescents with well-adjusted behaviour or occasional minor delinquency were less likely to report drinking large amounts of alcohol or to the point of feeling drunk.

CONCLUSIONS

The present results further emphasize the importance of distinguishing between different offender groups when examining the relationship between delinquent behaviour and associated problems, such as excessive alcohol use.