Is handling unbalanced datasets for machine learning uplifts system performance?: A case of diabetic prediction.

BACKGROUND AND AIMS: Healthcare is a sensitive sector, and addressing the class imbalance in the healthcare domain is a time-consuming task for machine learning-based systems due to the vast amount of data. This study looks into the impact of socioeconomic disparities on the healthcare data of diabetic patients to make accurate disease predictions. METHODS: This study proposed a systematic approach of Closest Distance Ranking and Principal Component Analysis to deal with the unbalanced dataset. A typical machine learning technique was used to analyze the proposed approach. The data set of pregnant diabetic women is analysed for accurate detection. RESULTS: The results of the case are analysed using sensitivity, which demonstrates that the minority class's lack of information makes it impossible to forecast the results. On the other hand, the unbalanced dataset was treated using the proposed technique and evaluated with the machine learning algorithm which significantly increased the performance of the system. CONCLUSION: The performance of the machine learning-based system was significantly enhanced by the unbalanced dataset which was processed with the proposed technique and evaluated with the machine learning algorithm. For the first time, an unbalanced dataset was treated with a combination of Closest Distance Ranking and Principal Component Analysis.

Peripheral lymphocyte subset alteration in patients with COVID-19 having differential clinical manifestations.

Background & objectives: The COVID-19 disease profile in Indian patients has been found to be different from the Western world. Changes in lymphocyte compartment have been correlated with disease course, illness severity and clinical outcome. This study was aimed to assess the peripheral lymphocyte phenotype and subset distribution in patients with COVID-19 disease from India with differential clinical manifestations. Methods: Percentages of peripheral lymphocyte subsets were measured by flow cytometry in hospitalized asymptomatic (n=53), mild symptomatic (n=36), moderate and severe (n=30) patients with SARS-CoV-2 infection, recovered individuals (n=40) and uninfected controls (n=56) from Pune, Maharashtra, India. Results: Percentages of CD4+Th cells were significantly high in asymptomatic, mild symptomatic, moderate and severe patients and recovered individuals compared to controls. Percentages of Th memory (CD3+CD4+CD45RO+), Tc memory (CD3+CD8+CD45RO+) and B memory (CD19+CD27+) cells were significantly higher in the recovered group compared to both asymptomatic, mild symptomatic patient and uninfected control groups. NK cell (CD56+CD3-) percentages were comparable among moderate +severe patient and uninfected control groups. Interpretation & conclusions: The observed lower CD4+Th cells in moderate+severe group requiring oxygen support compared to asymptomatic+mild symptomatic group not requiring oxygen support could be indicative of poor prognosis. Higher Th memory, Tc memory and B memory cells in the recovered group compared to mild symptomatic patient groups might be markers of recovery from mild infection; however, it remains to be established if the persistence of any of these cells could be considered as a correlate of protection.

Incidence, Clinical Features, and Outcomes of Transient Tachypnea of the Newborn at a Tertiary Care Center in Western India.

Background Transient tachypnea of the newborn (TTN) is a self-limiting, benign condition leading to respiratory distress shortly after birth. It is among the leading cause of respiratory distress in term and late preterm neonates. The disease is transient and resolves by three to four days in most neonates. Objective The objective of this study was to study the incidence of TTN, its clinical features, predictors of outcomes and duration of hospital stay in these neonates suffering from it. Methods This was a prospective study done at a tertiary care center carried out between August 2019 to July 2021. The study subjects were late pre-term (34 to 36 weeks of gestation) and term neonates with respiratory distress who were admitted to the neonatal intensive care unit (NICU). The diagnosis was based on clinical features, radiological features, and clinical course in NICU. Results The total number of cases with TTN was 74. The incidence of TTN was 16 per 1000 live births. 63.5% were male, 75.7% were term births, 70.3% were born via lower section cesarean section (LSCS), and 66.2% were normal birth weight (≥2.5 kg) infants. A high incidence of TTN was found in late pre-term babies, babies born via LSCS, and male sex. None of the neonates required ventilatory support, either noninvasive or invasive. Conclusion Delivery by LSCS and male sex were risk factors for the development of TTN. The distress in TTN is usually mild to moderate, and in most cases, oxygen supplementation suffices. Higher Downes' score at presentation, low birth weight, preterm, and delivery by LSCS were found to be predictors for a longer duration of distress and thus the longer duration of NICU stay. Although severe complications for TTN have been reported in the literature, they are rare. Careful observation can decrease not only a lot of unnecessary investigations but also allow clinicians at secondary and primary centers to better care for neonates with TTN.

Pro-inflammatory CXCL-10, TNF-α, IL-1ß, and IL-6: biomarkers of SARS-CoV-2 infection.

Currently, the world is witnessing the pandemic of COVID-19, a disease caused by the novel coronavirus SARS-CoV-2. Reported differences in clinical manifestations and outcomes in SARS-CoV-2 infection could be attributed to factors such as virus replication, infiltration of inflammatory cells, and altered cytokine production. Virus-induced aberrant and excessive cytokine production has been linked to the morbidity and mortality of several viral infections. Using a Luminex platform, we investigated plasma cytokine and chemokine levels of 27 analytes from hospitalized asymptomatic (n = 39) and mildly symptomatic (n = 35) SARS-CoV-2-infected patients (in the early phase of infection), recovered individuals (45-60 days postinfection) (n = 40), and uninfected controls (n = 36) from the city of Pune located in the state of Maharashtra in India. Levels of the pro-inflammatory cytokines IL-1ß, IL-6, and TNF-α and the chemokine CXCL-10 were significantly higher, while those of the antiviral cytokines IFN-γ and IL-12 p70 were significantly lower in both asymptomatic and mildly symptomatic patients than in controls. Comparison among the patient categories revealed no difference in the levels of the cytokines/chemokines except for CXCL-10 being significantly higher and IL-17, IL-4, and VEGF being significantly lower in the mildly symptomatic patients. Interestingly, levels of all key analytes were significantly lower in recovered individuals than in those in both patient categories. Nevertheless, the level of CXCL10 was significantly higher in the recovered patients than in the controls, indicating that the immune system of SARS-CoV-2 patients may take a longer time to normalize. Our data suggest that IL-6, IL-1ß, TNF-α, CXCL-10, and reduced antiviral cytokines could be used as biomarkers of SARS-CoV-2 infection.

A cause of Sudden Cardiac Deaths on Autopsy Findings; a Four-Year Report.

INTRODUCTION: Incidence of sudden cardiac death (SCD) has been steadily increasing all over the world. While knowing the cause of SCD is one of the favorites of the physicians involved with these cases, it is very difficult and challenging task for the forensic physician. The present report is a prospective study regarding cause of SCDs on autopsy examination in four-year period, Bangalore, India. METHODS: The present prospective study is based on autopsy observations, carried out for four-year period from 2008 to 2011, and analyzed for cause of SCDs. The cases were chosen as per the definition of sudden death and autopsied. The material was divided into natural and unnatural groups. Finally, on histopathology, gross examination, hospital details, circumstantial, and police reports the cause of death was inferred. RESULTS: A total of 2449 autopsy was conducted of which 204 cases were due to SCD. The highest SCDs were reported in 50-60 years age group (62.24%; n-127), followed closely by the age group 60-69 (28.43%; n-58). Male to female ratio was around 10:1. The maximum number of deaths (n=78) was within few hours (6 hours) after the onset of signs and symptoms. In 24 (11.8%) cases major narrowing was noted in both the main coronaries, in 87 (42.6%) cases in the left anterior descending coronary artery (LAD), and in 18 (51.5%) cases in the right coronary artery (RCA). The major cardiac pathology resulting in sudden death was coronary artery disease (n-116; 56.86%) and myocardial infarction (n-104; 50.9%). most of the SCDs occurred in the place of residence (n-80; 39.2%) followed closely by death in hospital (n-49; 24.01%). CONCLUSION: Coronary occlusion was the major contributory cause of sudden death with cardiac origin and the highest number of deaths were reported in the age 50-59 years with male to female ratio of 10:1.

Trends in choosing place of delivery and assistance during delivery in Nanded district, Maharashtra, India.

Delivery in a medical institution promotes child survival and reduces the risk of maternal mortality. Many initiatives under the National Rural Health Mission (NRHM) focus on increasing the institutional deliveries. This study describes the trends in choosing place of delivery in Nanded district at the end of the first phase of the mission. Key informants were interviewed to document the initiatives under NRHM implemented in the district. A cross-sectional descriptive study was conducted in 30 villages selected using one stage cluster-sampling method. A house-to-house survey was conducted in June 2009. A set of structured open-ended questionnaire was used for interviewing all women who had delivered during January 2004-May 2009. The outcomes studied were place of delivery and assistance during delivery. Analysis was done by calculating chi-square test and odds ratio. Interventions to improve the quality of health services and healthcare-seeking behaviour were implemented successfully in the district. The proportion of institutional deliveries increased from 42% in 2004 to 69% in 2009. A significant increase was observed in the proportion of institutional deliveries [60% vs 45%; chi2 = 173.85, p < 0.05, odds ratio (OR) = 1.8 (95% confidence interval (CI) 1.65-1.97)] in the NRHM period compared to the pre-NRHM period. The deliveries in government institutions and in private institutions also showed a significant rise. The proportion of deliveries assisted by health personnel increased significantly during the NRHM period [62% vs 49%; chi2 = 149.39; p < 0.05, OR = 1.73, 95% CI 1.58-1.89]. However, less than 10% of the deliveries in the home (range 2-9%) were assisted by health personnel throughout the study period. There was a wide geographic variation in place of delivery among the study villages. The results showed a significant increase in the proportion of institutional deliveries and deliveries assisted by health personnel in the NRHM period. Since a less proportion of deliveries in the home is conducted by health personnel, the focus should be on increasing the institutional deliveries. Special and innovative interventions should be implemented in the villages with a less proportion of institutional deliveries.

TAGmapper: a web-based tool for mapping SAGE tags.

Serial Analysis of Gene Expression (SAGE) is an important means of obtaining quantitative information about expression of genes in different samples. Short SAGE tags are 10 nucleotides long and often contain enough information to uniquely identify the gene(s) corresponding to the tag. We have observed, however, that the currently available resources are not adequate for accurate mapping of all SAGE tags to genes. Here, we describe development of a web-based tool called TAGmapper (http://tagmapper.ibioinformatics.org), which provides a comprehensive and accurate mapping of SAGE tags to genes. We were able to map SAGE tags accurately in several instances where two other popular resources, SAGEmap (http://www.ncbi.nlm.nih.gov/projects/SAGE/) and SAGE Genie (http://cgap.nci.nih.gov/SAGE), provided incorrect or no assignment of tags to genes. Finally, we experimentally determined the expression of a subset of genes assigned by TAGmapper using DNA microarrays and/or quantitative PCR to confirm the reliability of the gene mappings. We anticipate that TAGmapper will be a useful tool in functional genomic approaches by providing accurate identification of genes in SAGE experiments.

Plasma Proteome Database as a resource for proteomics research.

Plasma is one of the best studied compartments in the human body and serves as an ideal body fluid for the diagnosis of diseases. This report provides a detailed functional annotation of all the plasma proteins identified to date. In all, gene products encoded by 3778 distinct genes were annotated based on proteins previously published in the literature as plasma proteins and the identification of multiple peptides from proteins under HUPO's Plasma Proteome Project. Our analysis revealed that 51% of these genes encoded more than one protein isoform. All single nucleotide polymorphisms involving protein-coding regions were mapped onto the protein sequences. We found a number of examples of isoform-specific subcellular localization as well as tissue expression. This database is an attempt at comprehensive annotation of a complex subproteome and is available on the web at http://www.plasmaproteomedatabase.org.

Differentially expressed genes in pancreatic ductal adenocarcinomas identified through serial analysis of gene expression.

Serial analysis of gene expression (SAGE) is a powerful tool for the discovery of novel tumor markers. The publicly available online SAGE libraries of normal and neoplastic tissues (http://www.ncbi.nlm.nih.gov/SAGE/) have recently been expanded; in addition, a more complete annotation of the human genome and better biocomputational techniques have substantially improved the assignment of differentially expressed SAGE "tags" to human genes. These improvements have provided us with an opportunity to re-evaluate global gene expression in pancreatic cancer using existing SAGE libraries. SAGE libraries generated from six pancreatic cancers were compared to SAGE libraries generated from 11 non-neoplastic tissues. Compared to normal tissue libraries, we identified 453 SAGE tags as differentially expressed in pancreatic cancer, including 395 that mapped to known genes and 58 "uncharacterized" tags. Of the 395 SAGE tags assigned to known genes, 223 were overexpressed in pancreatic cancer, and 172 were underexpressed. In order to map the 58 uncharacterized differentially expressed SAGE tags to genes, we used a newly developed resource called TAGmapper (http://tagmapper.ibioinformatics.org), to identify 16 additional differentially expressed genes. The differential expression of seven genes, involved in multiple cellular processes such as signal transduction (MIC-1), differentiation (DMBT1 and Neugrin), immune response (CD74), inflammation (CXCL2), cell cycle (CEB1) and enzymatic activity (Kallikrein 6), was confirmed by either immunohistochemical labeling of tissue microarrays (Kallikrein 6, CD74 and DMBT1) or by RT-PCR (CEB1, Neugrin, MIC1 and CXCL2). Of note, Neugrin was one of the genes whose previously uncharacterized SAGE tag was correctly assigned using TAGmapper, validating the utility of this program. Novel differentially expressed genes in a cancer type can be identified by revisiting updated and expanded SAGE databases. TAGmapper should prove to be a powerful tool for the discovery of novel tumor markers through assignment of uncharacterized SAGE tags.