Predicting Unreported Micronutrients From Food Labels: Machine Learning Approach.

BACKGROUND: Micronutrient deficiencies represent a major global health issue, with over 2 billion individuals experiencing deficiencies in essential vitamins and minerals. Food labels provide consumers with information regarding the nutritional content of food items and have been identified as a potential tool for improving diets. However, due to governmental regulations and the physical limitations of the labels, food labels often lack comprehensive information about the vitamins and minerals present in foods. As a result, information about most of the micronutrients is absent from existing food labels. OBJECTIVE: This paper aims to examine the possibility of using machine learning algorithms to predict unreported micronutrients such as vitamin A (retinol), vitamin C, vitamin B1 (thiamin), vitamin B2 (riboflavin), vitamin B3 (niacin), vitamin B6, vitamin B12, vitamin E (alpha-tocopherol), vitamin K, and minerals such as magnesium, zinc, phosphorus, selenium, manganese, and copper from nutrition information provided on existing food labels. If unreported micronutrients can be predicted with acceptable accuracies from existing food labels using machine learning predictive models, such models can be integrated into mobile apps to provide consumers with additional micronutrient information about foods and help them make more informed diet decisions. METHODS: Data from the Food and Nutrient Database for Dietary Studies (FNDDS) data set, representing a total of 5624 foods, were used to train a diverse set of machine learning classification and regression algorithms to predict unreported vitamins and minerals from existing food label data. For each model, hyperparameters were adjusted, and the models were evaluated using repeated cross-validation to ensure that the reported results were not subject to overfitting. RESULTS: According to the results, while predicting the exact quantity of vitamins and minerals is shown to be challenging, with regression R2 varying in a wide range from 0.28 (for magnesium) to 0.92 (for manganese), the classification models can accurately predict the category ("low," "medium," or "high") level of all minerals and vitamins with accuracies exceeding 0.80. The highest classification accuracies for specific micronutrients are achieved for vitamin B12 (0.94) and phosphorus (0.94), while the lowest are for vitamin E (0.81) and selenium (0.83). CONCLUSIONS: This study demonstrates the feasibility of predicting unreported micronutrients from existing food labels using machine learning algorithms. The results show that the approach has the potential to significantly improve consumer knowledge about the micronutrient content of the foods they consume. Integrating these predictive models into mobile apps can enhance their accessibility and engagement with consumers. The implications of this research for public health are noteworthy, underscoring the potential of technology to augment consumers' understanding of the micronutrient content of their diets while also facilitating the tracking of food intake and providing personalized recommendations based on the micronutrient content and individual preferences.

Association Between What People Learned About COVID-19 Using Web Searches and Their Behavior Toward Public Health Guidelines: Empirical Infodemiology Study.

BACKGROUND: The use of the internet and web-based platforms to obtain public health information and manage health-related issues has become widespread in this digital age. The practice is so pervasive that the first reaction to obtaining health information is to "Google it." As SARS-CoV-2 broke out in Wuhan, China, in December 2019 and quickly spread worldwide, people flocked to the internet to learn about the novel coronavirus and the disease, COVID-19. Lagging responses by governments and public health agencies to prioritize the dissemination of information about the coronavirus outbreak through the internet and the World Wide Web and to build trust gave room for others to quickly populate social media, online blogs, news outlets, and websites with misinformation and conspiracy theories about the COVID-19 pandemic, resulting in people's deviant behaviors toward public health safety measures. OBJECTIVE: The goals of this study were to determine what people learned about the COVID-19 pandemic through web searches, examine any association between what people learned about COVID-19 and behavior toward public health guidelines, and analyze the impact of misinformation and conspiracy theories about the COVID-19 pandemic on people's behavior toward public health measures. METHODS: This infodemiology study used Google Trends' worldwide search index, covering the first 6 months after the SARS-CoV-2 outbreak (January 1 to June 30, 2020) when the public scrambled for information about the pandemic. Data analysis employed statistical trends, correlation and regression, principal component analysis (PCA), and predictive models. RESULTS: The PCA identified two latent variables comprising past coronavirus epidemics (pastCoVepidemics: keywords that address previous epidemics) and the ongoing COVID-19 pandemic (presCoVpandemic: keywords that explain the ongoing pandemic). Both principal components were used significantly to learn about SARS-CoV-2 and COVID-19 and explained 88.78% of the variability. Three principal components fuelled misinformation about COVID-19: misinformation (keywords "biological weapon," "virus hoax," "common cold," "COVID-19 hoax," and "China virus"), conspiracy theory 1 (ConspTheory1; keyword "5G" or "@5G"), and conspiracy theory 2 (ConspTheory2; keyword "ingest bleach"). These principal components explained 84.85% of the variability. The principal components represent two measurements of public health safety guidelines-public health measures 1 (PubHealthMes1; keywords "social distancing," "wash hands," "isolation," and "quarantine") and public health measures 2 (PubHealthMes2; keyword "wear mask")-which explained 84.7% of the variability. Based on the PCA results and the log-linear and predictive models, ConspTheory1 (keyword "@5G") was identified as a predictor of people's behavior toward public health measures (PubHealthMes2). Although correlations of misinformation (keywords "COVID-19," "hoax," "virus hoax," "common cold," and more) and ConspTheory2 (keyword "ingest bleach") with PubHealthMes1 (keywords "social distancing," "hand wash," "isolation," and more) were r=0.83 and r=-0.11, respectively, neither was statistically significant (P=.27 and P=.13, respectively). CONCLUSIONS: Several studies focused on the impacts of social media and related platforms on the spreading of misinformation and conspiracy theories. This study provides the first empirical evidence to the mainly anecdotal discourse on the use of web searches to learn about SARS-CoV-2 and COVID-19.

Lactate/albumin ratio: An early prognostic marker in critically ill patients.

OBJECTIVES: We investigate the clinical utility of the lactate/albumin (L/A) ratio as an early prognostic marker of ICU mortality in a large cohort of unselected critically ill patients. METHODS: A retrospective single-center study using data from the Multiparameter Intelligent Monitoring Intensive Care III (MIMIC-III) database collected between 2001 and 2012. We screened adult patients (age ≥ 15) with measured lactate and albumin on the first day of ICU stay to evaluate the prognostic performance of the lactate and lactate/albumin (L/A) ratio for ICU mortality prediction. RESULTS: The overall ICU mortality in the 6414 eligible ICU patients was 16.4%. L/A showed a receiver-operating characteristics area under the curve (ROC-AUC) value of 0.69 (95% CI: 0.67, 0.70) to predict ICU mortality, higher than lactate 0.67 (95%CI: 0.65, 0.69). Regardless of the lactate level, L/A yielded better ROC-AUC compared to the lactate level [normal lactate (<2.0 mmol/L): 0.63 vs 0.60; intermediate lactate (2.0 mmol/L ≤ lactate <4.0 mmol/L): 0.58 vs 0.56; high lactate (≥4.0 mmol/L): 0.67 vs 0.66]. L/A was a better prognostic marker for ICU mortality in patients with decreased lactate elimination [hepatic dysfunction: 0.72 vs 0.70; renal dysfunction 0.70 vs 0.68]. The L/A ratio ROC-AUC was better in patients with sepsis (0.68 vs 0.66) and those who developed severe sepsis or septic shock (0.68 vs 0.66). CONCLUSIONS: The performance of L/A and lactate were equivalent in predicting ICU mortality and can be used as early prognostic markers for ICU patients with different initial lactate level and the presence of hepatic or renal dysfunction.

Effect of single nucleotide polymorphisms in SEPS1 and SEPP1 on expression in the protein level in metabolic syndrome in subjects with cardiovascular disease.

Metabolic syndrome (MetS) results from the interaction between environmental and genetic factors. Several previous studies considered the role of selenium in developing MetS. Two selenoproteins, selenoprotein S (SelS), and the Selenoprotein P (SePP) play an important role in antioxidative defense and therefore susceptibility to MetS. The involvement of SNPs in SEPP1 and SEPS1 have not been studied in MetS subjects. This study aims to investigate the association between the risk of MetS and four polymorphisms SEPS1 (rs28665122, rs4965373), SEPP1 (rs7579, rs3877899) in an Iranian population. The sample of this case-control study consisted of 132 Iranian patients with cardiovascular disease (71 MetS and 65 non-MetS subjects) from December 2015 to March 2016. Demographic data, medical history, and para-clinical were measured, and Taqman probes were used for allelic discrimination. The level of the SelS and the SePP were measured by the ELIZA method. No significant differences were found in the genotype frequencies of SEPS1 (rs4965373, rs28665122), SEPP1 (rs7579, rs3877899) in patients with MetS and the non-MetS group. The mean of SelS in MetS subjects with SEPS1 (rs4965373) GG genotype is significantly lower than the non-MetS group (4496.99 ± 3688.5 vs. 6148.6 ± 1127.0, P = 0.009). The mean of SePP in MetS subjects with SEPP1 (rs3877899) GG genotype is significantly lower than the non-MetS group (40.73 ± 8.44 vs.83.91 ± 21.33, P = 0.002). The mean of SePP in MetS subjects with SEPP1 (rs7579) GG genotype is lower than the non-MetS group (55.52 ± 16.7 vs. 109.48 ± 29.78, P = 0.01). In summary, the results of this study does not indicate significant differences in the SEPP1 (rs7579, rs3877899) and SEPS1 (rs4965373, rs28665122) genotypes between MetS and non-MetS subjects. However, the results show that the mean of expression of SelS and SePP decreased in the subjects with SEPP1 (rs7579) GG and SEPP1 (rs3877899) GG.

Association of expression of selenoprotein P in mRNA and protein levels with metabolic syndrome in subjects with cardiovascular disease: Results of the Selenegene study.

BACKGROUND: Selenoprotein P (SeP) is involved in transporting selenium from the liver to target tissues. Because SeP confers protection against disease by reducing chronic oxidative stress, the present study aimed to assess the level of SeP in the serum of patients with metabolic syndrome (MetS) with a history of cardiovascular disease (CVD). METHODS: A cross-sectional study was conducted in 63 and 71 subjects with and without MetS in the presence of documented CVD. All demographic, anthropometric and cardiometabolic variables (lipids, blood glucose, blood pressure) were assessed. Lifestyle-related factors and personal history and familial CVD risk factors were recorded. The expression of SELP in mRNA and protein levels in the serum was measured, and MetS was determined using ATPIII criteria. Binary logistic regression analysis demonstrated MetS and SeP to be dependent and independent variables, respectively. RESULTS: Mean of systolic and diastolic blood pressure, triglyceride, high-density lipoprotein-cholesterol, fasting blood sugar, body mass index and waist circumference were higher among subjects with MetS (p = 0.05). The mean of selenium was higher among subjects with MetS, whereas the mean of SeP was lower among subjects with MetS (p < 0.001). In the unadjusted model, the SeP had decreased odds for MetS [odds ratio (OR) = 0.995; 95% confidence interval (CI) = 0.989-1.00] (p < 0.04). Furthermore, the association between MetS and SeP levels remained marginally significant even after adjusting for potential confounders such as age, gender, family history, smoking status and nutrition. SeP and waist circumference show a significant relationship (OR =0.995; 95% CI = 0.990-1.00) (p < 0.033). CONCLUSIONS: We have demonstrated a significant decrease in circulating SeP levels according to MetS status in patients with documented cardiovascular disease.

Predicting Sociodemographic Attributes from Mobile Usage Patterns: Applications and Privacy Implications.

When users interact with their mobile devices, they leave behind unique digital footprints that can be viewed as predictive proxies that reveal an array of users' characteristics, including their demographics. Predicting users' demographics based on mobile usage can provide significant benefits for service providers and users, including improving customer targeting, service personalization, and market research efforts. This study uses machine learning algorithms and mobile usage data from 235 demographically diverse users to examine the accuracy of predicting their sociodemographic attributes (age, gender, income, and education) from mobile usage metadata, filling the gap in the current literature by quantifying the predictive power of each attribute and discussing the practical applications and privacy implications. According to the results, gender can be most accurately predicted (balanced accuracy = 0.862) from mobile usage footprints, whereas predicting users' education level is more challenging (balanced accuracy = 0.719). Moreover, the classification models were able to classify users based on whether their age or income was above or below a certain threshold with acceptable accuracy. The study also presents the practical applications of inferring demographic attributes from mobile usage data and discusses the implications of the findings, such as privacy and discrimination risks, from the perspectives of different stakeholders.

The incidence and outcome of severe hyperlactatemia in critically ill patients.

This study aimed to assess the incidence, persistence, and associated mortality of severe hyperlactatemia in a large cohort of unselected critically ill patients. Also, we evaluated the association between 12 h lactate clearance, the timing of severe hyperlactatemia, and the maximum lactate levels with ICU mortality. In this retrospective, single-center study, we used data from the Multiparameter Intelligent Monitoring in Intensive Care III (MIMIC-III) database. Data extracted to screen 23,598 ICU patients for severe hyperlactatemia. A total of 23,598 critically ill patients were eligible for this study. Overall, ICU mortality in the 23,598 ICU patients was 12.1%. Of these, 760 patients had lactate concentration [Formula: see text] 10 mmol/L and ICU mortality in this group was 65%. Our findings confirm the association between hyperlactatemia and ICU mortality [odds ratio 1.42 (95% CI 1.35; 1.49; P < 0.001)]. Data for 12 h lactate clearance was available for 443 patients (276 nonsurvivable vs. 167 survival). 12 h lactate clearance yielded a high area under the curve (AUC) of 0.78, (95% CI 0.74 and 0.83). Severe hyperlactatemia is associated with extremely high ICU mortality in a heterogeneous ICU population. Lactate derived variables (the timing and persistence of severe hyperlactatemia, maximum level, and 12 h clearance) are shown to be associated with ICU mortality in patients with severe hyperlactatemia. Our results suggest that maximum lactate level and 12 h lactate clearance were clinically useful prognostic parameters for patients with severe hyperlactatemia.

Depression screening using mobile phone usage metadata: a machine learning approach.

OBJECTIVE: Depression is currently the second most significant contributor to non-fatal disease burdens globally. While it is treatable, depression remains undiagnosed in many cases. As mobile phones have now become an integral part of daily life, this study examines the possibility of screening for depressive symptoms continuously based on patients' mobile usage patterns. MATERIALS AND METHODS: 412 research participants reported a range of their mobile usage statistics. Beck Depression Inventory-2nd ed (BDI-II) was used to measure the severity of depression among participants. A wide array of machine learning classification algorithms was trained to detect participants with depression symptoms (ie, BDI-II score ≥ 14). The relative importance of individual variables was additionally quantified. RESULTS: Participants with depression were found to have fewer saved contacts on their devices, spend more time on their mobile devices to make and receive fewer and shorter calls, and send more text messages than participants without depression. The best model was a random forest classifier, which had an out-of-sample balanced accuracy of 0.768. The balanced accuracy increased to 0.811 when participants' age and gender were included. DISCUSSIONS/CONCLUSION: The significant predictive power of mobile usage attributes implies that, by collecting mobile usage statistics, mental health mobile applications can continuously screen for depressive symptoms for initial diagnosis or for monitoring the progress of ongoing treatments. Moreover, the input variables used in this study were aggregated mobile usage metadata attributes, which has low privacy sensitivity making it more likely for patients to grant required application permissions.