A Risk Prediction Model for Invasive Fungal Disease in Critically Ill Patients in the Intensive Care Unit

Predicting the risk of invasive fungal infections in ICU sepsis population: the AMI risk assessment tool

BACKGROUND

Invasive fungal infections (IFI) represent a significant contributor to mortality among sepsis patients in the Intensive Care Unit (ICU). Early diagnosis of IFI is challenging, and currently, there are no predictive tools for identifying sepsis patients who may develop IFI. Our study aims to develop a predictive scoring system to assess the risk of IFI in patients with sepsis admitted to the ICU.

METHODS

A retrospective collection of data from a total of 549 patients was conducted. Data-driven, clinically knowledge-driven, and decision tree models were used to identify predictive variables for risk of IFI in ICU patients with sepsis. Demographic data, vital signs, laboratory values, comorbidities, medication use, and clinical outcomes were all collected. The optimal model was selected based on model performance and clinical utility to establish a risk score.

RESULTS

Among adult patients with sepsis admitted to the ICU, 127 patients (23.1%) developed IFI. The final data-driven model included four predictive factors, the clinically knowledge-driven model included three predictive factors, and the decision tree model included two. Based on the good performance and clinical utility of the clinically knowledge-driven model, it was chosen as the optimal risk scoring model (C-statistics: 0.79 (95% confidence interval (CI): 0.75-0.83); Hosmer-Lemeshow (H-L) test P = 0.884). The ICU sepsis patient invasive fungal infection risk (AMI) score, created based on the clinically knowledge-driven model, includes mechanical ventilation, application of immunosuppressants, and the types of antibiotics used. The C-statistics for this risk score was 0.79 (95% CI:0.75-0.84) with good calibration (H-L test P = 0.992 and see calibration curve: Fig. 2). Moreover, in terms of clinical utility, the decision curve analysis for AMI showed a favorable net benefit.

CONCLUSIONS

The application of the AMI score can effectively distinguish whether ICU sepsis patients will develop IFI, which is beneficial for clinicians to formulate targeted and timely preventive and treatment measures based on the risk of IFI.

INTERPRETABLE MACHINE LEARNING FOR PREDICTING RISK OF INVASIVE FUNGAL INFECTION IN CRITICALLY ILL PATIENTS IN THE INTENSIVE CARE UNIT: A RETROSPECTIVE COHORT STUDY BASED ON MIMIC-IV DATABASE

ABSTRACT

The delayed diagnosis of invasive fungal infection (IFI) is highly correlated with poor prognosis in patients. Early identification of high-risk patients with invasive fungal infections and timely implementation of targeted measures is beneficial for patients. The objective of this study was to develop a machine learning-based predictive model for invasive fungal infection in patients during their intensive care unit (ICU) stay. Retrospective data was extracted from adult patients in the MIMIC-IV database who spent a minimum of 48 h in the ICU. Feature selection was performed using LASSO regression, and the dataset was balanced using the BL-SMOTE approach. Predictive models were built using six machine learning algorithms. The Shapley additive explanation algorithm was used to assess the impact of various clinical features in the optimal model, enhancing interpretability. The study included 26,346 ICU patients, of whom 379 (1.44%) were diagnosed with invasive fungal infection. The predictive model was developed using 20 risk factors, and the dataset was balanced using the borderline-SMOTE (BL-SMOTE) algorithm. The BL-SMOTE random forest model demonstrated the highest predictive performance (area under curve = 0.88, 95% CI = 0.84-0.91). Shapley additive explanation analysis revealed that the three most influential clinical features in the BL-SMOTE random forest model were dialysis treatment, APSIII scores, and liver disease. The machine learning model provides a reliable tool for predicting the occurrence of IFI in ICU patients. The BL-SMOTE random forest model, based on 20 risk factors, exhibited superior predictive performance and can assist clinicians in early assessment of IFI occurrence in ICU patients. Importance: Invasive fungal infections are characterized by high incidence and high mortality rates characteristics. In this study, we developed a clinical prediction model for invasive fungal infections in critically ill patients based on machine learning algorithms. The results show that the machine learning model based on 20 clinical features has good predictive value.

Breaking barriers in Candida spp. detection with Electronic Noses and artificial intelligence

The timely and accurate diagnosis of candidemia, a severe bloodstream infection caused by Candida spp., remains challenging in clinical practice. Blood culture, the current gold standard technique, suffers from lengthy turnaround times and limited sensitivity. To address these limitations, we propose a novel approach utilizing an Electronic Nose (E-nose) combined with Time Series-based classification techniques to analyze and identify Candida spp. rapidly, using culture species of C. albicans, C.kodamaea ohmeri, C. glabrara, C. haemulonii, C. parapsilosis and C. krusei as control samples. This innovative method not only enhances diagnostic accuracy and reduces decision time for healthcare professionals in selecting appropriate treatments but also offers the potential for expanded usage and cost reduction due to the E-nose's low production costs. Our proof-of-concept experimental results, carried out with culture samples, demonstrate promising outcomes, with the Inception Time classifier achieving an impressive average accuracy of 97.46% during the test phase. This paper presents a groundbreaking advancement in the field, empowering medical practitioners with an efficient and reliable tool for early and precise identification of candidemia, ultimately leading to improved patient outcomes.

Synthesis of vanillin derivatives with 1,2,3-triazole fragments and evaluation of their fungicide and fungistatic activities

Vanillin is the main component of natural vanilla extract and is responsible for its flavoring properties. Besides its well-known applications as an additive in food and cosmetics, it has also been reported that vanillin can inhibit fungi of clinical interest, such as Candida spp., Cryptococcus spp., Aspergillus spp., as well as dermatophytes. Thus, the present work approaches the synthesis of a series of vanillin derivatives with 1,2,3-triazole fragments and the evaluation of their antifungal activities against Candida albicans, Candida glabrata, Candida parapsilosis, Candida tropicalis, Cryptococcus neoformans, Cryptococcus gattii, Trichophyton rubrum, and Trichophyton interdigitale strains. Twenty-two vanillin derivatives were obtained, with yields in the range of 60%-91%, from copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC) click reaction between two terminal alkynes prepared from vanillin and different benzyl azides. In general, the evaluated compounds showed moderate activity against the microorganisms tested, with minimum inhibitory concentration (MIC) values ranging from 32 to >512 µg mL^-1 . Except for compound 3b against the C. gattii R265 strain, all vanillin derivatives showed fungicidal activity for the yeasts tested. The predicted physicochemical and ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties for the compounds indicated favorable profiles for drug development. In addition, a four-dimensional structure-activity relationship (4D-SAR) analysis was carried out and provided useful insights concerning the structures of the compounds and their biological profile. Finally, molecular docking calculations showed that all compounds bind favorably at the lanosterol 14α-demethylase enzyme active site with binding energies ranging from -9.1 to -12.2 kcal/mol.

Challenges and Opportunities in Understanding Genetics of Fungal Diseases: Towards a Functional Genomics Approach

Infectious diseases are a leading cause of morbidity and mortality worldwide, and human pathogens have long been recognized as one of the main sources of evolutionary pressure, resulting in a high variable genetic background in immune-related genes. The study of the genetic contribution to infectious diseases has undergone tremendous advances over the last decades. Here, focusing on genetic predisposition to fungal diseases, we provide an overview of the available approaches for studying human genetic susceptibility to infections, reviewing current methodological and practical limitations. We describe how the classical methods available, such as family-based studies and candidate gene studies, have contributed to the discovery of crucial susceptibility factors for fungal infections. We will also discuss the contribution of novel unbiased approaches to the field, highlighting their success but also their limitations for the fungal immunology field. Finally, we show how a systems genomics approach can overcome those limitations and can lead to efficient prioritization and identification of genes and pathways with a critical role in susceptibility to fungal diseases. This knowledge will help to stratify at-risk patient groups and, subsequently, develop early appropriate prophylactic and treatment strategies.

Data quality assessment and subsampling strategies to correct distributional bias in prevalence studies

Background

Healthcare-associated infections (HAIs) represent a major Public Health issue. Hospital-based prevalence studies are a common tool of HAI surveillance, but data quality problems and non-representativeness can undermine their reliability.

Methods

This study proposes three algorithms that, given a convenience sample and variables relevant for the outcome of the study, select a subsample with specific distributional characteristics, boosting either representativeness (Probability and Distance procedures) or risk factors’ balance (Uniformity procedure). A “Quality Score” (QS) was also developed to grade sampled units according to data completeness and reliability.

The methodologies were evaluated through bootstrapping on a convenience sample of 135 hospitals collected during the 2016 Italian Point Prevalence Survey (PPS) on HAIs.

Results

The QS highlighted wide variations in data quality among hospitals (median QS 52.9 points, range 7.98–628, lower meaning better quality), with most problems ascribable to ward and hospital-related data reporting. Both Distance and Probability procedures produced subsamples with lower distributional bias (Log-likelihood score increased from 7.3 to 29 points). The Uniformity procedure increased the homogeneity of the sample characteristics (e.g., − 58.4% in geographical variability).

The procedures selected hospitals with higher data quality, especially the Probability procedure (lower QS in 100% of bootstrap simulations). The Distance procedure produced lower HAI prevalence estimates (6.98% compared to 7.44% in the convenience sample), more in line with the European median.

Conclusions

The QS and the subsampling procedures proposed in this study could represent effective tools to improve the quality of prevalence studies, decreasing the biases that can arise due to non-probabilistic sample collection.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01277-y.

Developement of health risk rating scale for indoor airborne fungal exposure

This paper aims to quantify airborne fungal load in air-conditioned rooms and develop a health risk rating scale for different indoor environments. Five sampling locations in Kolkata frequented by a heterogeneous human population, containing various types of fungal growth-promoting substances (FGPS) like old documents, food items, waste hair, etc. were chosen as sampling locations where an Andersen Two-Stage Cascade Impactor was ran using Rose Bengal agar and Potato Dextrose agar media plates. Total spore load (CFU/m³), species diversity, species dominance, human exposure time, susceptible age and FGPS were considered the risk factors for this study. A risk rating scale was developed after evaluating the relative importance of these different factors in relation to human health. The most dominant genera were Aspergillus, followed by Penicillium. Maximum CFU was observed at library, followed by computer room.