: Artificial Intelligence techniques are widely used for medical purposes nowadays. One of the crucial applications is cancer detection. Due to the sensitivity of such applications, medical workers and patients interacting with the system must get a reliable, transparent, and explainable output. Therefore, this paper examines the interpretability and explainability of the Logistic Regression Model (LRM) for breast cancer detection. We analyze the accuracy and transparency of the LRM model. Additionally, we propose an NLP-based interface with a model interpretability summary and a contrastive explanation for users. Together with textual explanations, we provide a visual aid for medical practitioners to understand the decision-making process better.

Social media has opened the gates for collecting big data that can be used to monitor epidemic trends in real time. We evaluate whether Watson NLP service can be used to reliably predict infectious disease such as influenza-like illness (ILI) outbreaks using Twitter data during the period of the main influenza season. Watson’s performance is evaluated by computing Pearson correlation between the number of tweets classified by Watson as ILI and the number of ILI occurrences recovered from traditional epidemic surveillance system of the Centers for Disease Control and Prevention (CDC). Achieved correlation was 0.55. Furthermore, a 12 week discrepancy was found between peak occurrences of ILI predicted by Watson and CDC reported data. Additionally, we developed a scoring method for ILI prediction from Twitter posts using a simple formula with the ability to predict ILI two weeks ahead of CDC reported ILI data. The method uses Watson’s sentiment and emotion scores together with identified ILI features to analyze influenza-related posts in real time. Due to Watson's high computational costs of sentiment and emotion analysis, we tested if machine learning approach can be used to predict influenza using only identified ILI keywords as influenza predictors. All three evaluated methods (Random Forest, Logistic Regression, K-NN), achieved overall accuracy of ~68.2% and 97.5% respectively, when Watson and the developed formula are used as medical experts. The obtained results suggest that data found within social media can be used to supplement the traditional surveillance of influenza outbreaks with the help of intelligent computations.

Abstract Lavender and immortelle essential oils (EOs) are widely used to treat a spectrum of human conditions. The aim of this study was to investigate cyto/genotoxic effects of lavender and immortelle EOs using plant cells (Allium cepa) and human lymphocytes, as well as their antimicrobial potential using nine strains of bacteria and fungi. Our results for lavender and immortelle EOs showed that the frequency of chromosome aberrations (CAs) was increased in comparison with controls. For both oils, increased frequency of apoptosis for all concentrations, as well as the frequency of necrosis (0.10/0.30 µl/ml for lavender/immortelle, respectively) was demonstrated. In human lymphocytes, differences for minute fragments between immortelle oil (0.10 µl/ml) and controls were observed. Increased frequency of apoptosis was detected for immortelle oil (0.20 µl/ml), while both oils (0.20; 0.30 µl/ml lavender, and immortelle at all concentrations) induced higher frequency of necrosis in comparison with controls. Lavender EO was effective against all tested Gram-positive and Gram-negative bacteria, while immortelle EO inhibited only Gram-positive bacteria. Both oils exhibited antifungal effect. Our results demonstrated that lavender and immortelle EOs showed cyto/genotoxic effects in both, plant and human cells, as well as antimicrobial properties. Further studies are needed to strengthen these findings.

Abstract Objective: We assessed the ability of the Prime Diet Quality Score (PDQS) to predict mortality in the US population and compared its predictiveness with that of the Healthy Eating Index-2015 (HEI-2015). Design: PDQS and HEI-2015 scores were derived using two 24-h recalls and converted to quintiles. Mortality data were obtained from the 2015 Public-Use Linked Mortality File. Associations between diet quality and all-cause mortality were evaluated using multivariable Cox proportional hazards models, and predictive performance of the two metrics was compared using a Wald test of equality of coefficients with both scores in a single model. Finally, we evaluated associations between individual metric components and mortality. Setting: A prospective analysis of the US National Health and Nutrition Examination Survey (NHANES) data. Participants: Five-thousand five hundred and twenty-five participants from three survey cycles (2003–2008) in the NHANES aged 40 years and over. Results: Over the 51 248 person-years of follow-up (mean: 9·2 years), 767 deaths were recorded. In multivariable models, hazard ratios between the highest and lowest quintiles of diet quality scores were 0·70 (95 % CI 0·51, 0·96, Ptrend = 0·03) for the PDQS and 0·77 (95 % CI 0·57, 1·03, Ptrend = 0·20) for the HEI-2015. The PDQS and HEI-2015 were similarly good predictors of total mortality (Pdifference = 0·88). Conclusion: Among US adults, better diet quality measured by the PDQS was associated with reduced risk of all-cause mortality. Given that the PDQS is simpler to calculate than the HEI-2015, it should be evaluated further for use as a diet quality metric globally.

Between March 5 and July 25, 2020, the total number of SARS-CoV-2 confirmed cases in Bosnia and Herzegovina (BH) was 10 090 corresponding to a cumulative incidence rate of 285.7 per 100 000 population. Demographic and clinical information on all the cases along with exposure and contact information was collected using a standardized case report form. In suspected SARS-CoV-2 cases, respiratory specimens were collected and tested by real-time reverse-transcriptase polymerase chain reaction (rRT-PCR) assay. The dynamic of the outbreak was summarized using epidemiological curves, instantaneous reproduction number Rt and interactive choropleth maps for geographical distribution and spread. The rate of hospitalization was 14.0% (790/5646) in Federation of Bosnia and Herzegovina (FBH) and 6.2% (267/4299) in Republic of Srpska (RS). The death rate was 2.2% (122/5646) in FBH and 3.6% in the RS (155/4299). After the authorities lifted mandatory quarantine restrictions, the basic reproduction number increased from 1.13 on May, the 20th to 1.72 on May the 31st. The outbreak concerns both entities, Federation of Bosnia and Herzegovina and Republika Srpska, and it is more pronounced in those aged 20-44 years. It is important to develop the communication and emergency plan for the SARS-CoV-2 outbreak in BH, including the mechanisms to allow the ongoing notification and updates at the national level.

Aim The damage caused by the COVID-19 pandemic has made the prevention of its further spread at the top of the list of priorities of many governments and state institutions responsible for health and civil protection around the world. This prevention implies an effective system of epidemiological surveillance and the application of timely and effective control measures. This research focuses on the application of techniques for modelling and geovisualization of epidemic data with the aim of simple and fast communication of analytical results via geoportal. Methods The paper describes the approach applied through the project of establishing the epidemiological location-intelligence system for monitoring the effectiveness of control measures in preventing the spread of COVID-19 in Bosnia and Herzegovina. Results Epidemic data were processed and the results related to spatio-temporal analysis of the infection spread were presented by compartmental epidemic model, reproduction number R, epi-curve diagrams as well as choropleth maps for different levels of administrative units. Geovisualization of epidemic data enabled the release of numerous information from described models and indicators, providing easier visual communication of the spread of the disease and better recognition of its trend. Conclusion The approach involves the simultaneous application of epidemic models and epidemic data geovisualization, which allows a simple and rapid evaluation of the epidemic situation and the effects of control measures. This contributes to more informative decision-making related to control measures by suggesting their selective application at the local level.

Purpose: To examine the effect of an opt-out default recruitment strategy compared to a conventional opt-in strategy on enrollment and adherence to a behavioral intervention for poorly controlled diabetic patients. Design: Randomized controlled trial. Setting: University of Pennsylvania primary care practices. Participants: Participants of this trial included those with (1) age 18 to 80 years; (2) diabetes diagnosis; and (3) a measured hemoglobin A1c (HbA1c) greater than 8% in the past 12 months. Intervention: We randomized eligible patients into opt-in and opt-out arms prior to enrollment. Those in the opt-out arm received a letter stating that they were enrolled into a diabetes research study with the option to opt out, and those in the opt-in arm received a standard recruitment letter. Measures: Main end points include enrollment rate, defined as the proportion of participants who attended the baseline visit, and adherence to daily glycemic monitoring. Analysis: We powered our study to detect a 20% difference in adherence to device usage between arms and account for a 10% attrition rate. Results: Of the 569 eligible participants who received a recruitment letter, 496 were randomized to the opt-in arm and 73 to the opt-out arm. Enrollment rates were 38% in the opt-out arm and 13% in the opt-in arm (P < .001). Conclusions: Opt-out defaults, where clinically appropriate, could be a useful approach for increasing the generalizability of low-risk trials testing behavioral interventions in clinical settings.

Abstract This study aimed to examine the factors affecting feasibility of optimal and complete secondary cytoreductive surgery (SCRS) and to characterise the prognostic factors that correlate with improved survival in patients who underwent SCRS. This is a retrospective single-institutional cohort study of patients who underwent SCRS for recurrent epithelial ovarian cancer (EOC). One hundred and forty-eight patients met inclusion criteria. Platinum sensitivity was associated with complete cytoreduction at SCRS. Factors associated with suboptimal cytoreduction (SOC) were age >55 years, serous histology, largest tumour implant size >4 cm, and SOC at primary surgery. Overall survival analysis showed significantly longer survival with complete cytoreduction compared to optimal and SOC. Surgical outcome of SCRS was an independent predictor of survival regardless of the outcome of primary cytoreduction. Location of the largest implant, DFI and timing of chemotherapy also impact on survival.

With this prospective population-based study, we showed that infant neuromotor development, including muscle tone and senses, predicted internalizing problems but not externalizing problems through age 10 years. BACKGROUND: Research of adults and school-aged children suggest a neurodevelopmental basis for psychiatric disorders. We examined whether infant neuromotor development predicted internalizing and externalizing problems in young children. METHODS: In Generation R, a population-based cohort in the Netherlands (2002–2006), trained research assistants evaluated the neuromotor development of 4006 infants aged 2 to 5 months by using an adapted version of Touwen’s Neurodevelopmental Examination (tone, responses, and senses and other observations). We defined nonoptimal neuromotor development as scores in the highest tertile. Mothers and fathers rated their children’s behavior at ages 1.5, 3, 6, and 10 years with the Child Behavior Checklist (n = 3474, response: 86.7%). The associations were tested with generalized linear mixed models. RESULTS: Overall, neuromotor development predicted internalizing scores, but no association was observed with externalizing scores. Nonoptimal muscle tone was associated with higher internalizing scores (mothers’ report: β = .07; 95% confidence interval [CI]: 0.01 to 0.13; fathers’ report: β = .09, 95% CI: 0.00 to 0.16). In particular, nonoptimal low muscle tone was associated with higher internalizing scores (mothers’ report: β = .11; 95% CI: 0.05 to 0.18; fathers’ report: β = .13; 95% CI: 0.04 to 0.22). We also observed an association between senses and other observations with internalizing scores. There was no relationship between high muscle tone or reflexes and internalizing scores. CONCLUSIONS: Common emotional problems in childhood have a neurodevelopmental basis in infancy. Neuromotor assessment in infancy may help identify vulnerability to early internalizing symptoms and offer the opportunity for targeted interventions.

We specify identifying assumptions under which linear increments (LI) estimator can be used to estimate unconditional expectation for longitudinal data from a clinical trial in the presence of dropout. We show that these are analog conditions under which extended linear SWEEP estimator achieves unbiased estimation of the identical parameter in the same setting. Within a class of linear autoregressive models we specify how strategies implemented in LI and extended SWEEP relate to each other w.r.t. the conditional expectation of increments and outcomes respectively. We utilize conceptual overlap of these two methods to define a sensitivity analysis for both of them in presence of non-ignorable dropout. Interdependency of these two approaches offers a natural solution to a prominent problem of asynchronous association between outcome and dropout inevitably encountered in sensitivity analysis for dropout in longitudinal data. Validation of our approach is done on the data coming from a randomized, longitudinal trial of behavioral economic interventions to reduce CVD risk. We subsequently show that our approach to sensitivity analysis can be perceived as extension of the pattern mixture method defined by Daniels and Hogan in 2007. to longer sequences of observations. For T=3 we give the explicit expression for bias of our approach w.r.t. mentioned pattern mixture approach. We further show on a subset of the data from the same study that this bias does not invalidate our sensitivity analysis for LI when it comes to evaluating the robustness of findings under increasingly less ignorable dropout. Degree Type Dissertation Degree Name Doctor of Philosophy (PhD) Graduate Group Epidemiology & Biostatistics First Advisor Andrea B. Troxel