Deep continual learning for medical call incidents text classification under the presence of dataset shifts.

The aim of this work is to develop and evaluate a deep classifier that can effectively prioritize Emergency Medical Call Incidents (EMCI) according to their life-threatening level under the presence of dataset shifts. We utilized a dataset consisting of 1982746 independent EMCI instances obtained from the Health Services Department of the Region of Valencia (Spain), with a time span from 2009 to 2019 (excluding 2013). The dataset includes free text dispatcher observations recorded during the call, as well as a binary variable indicating whether the event was life-threatening. To evaluate the presence of dataset shifts, we examined prior probability shifts, covariate shifts, and concept shifts. Subsequently, we designed and implemented four deep Continual Learning (CL) strategies-cumulative learning, continual fine-tuning, experience replay, and synaptic intelligence-alongside three deep CL baselines-joint training, static approach, and single fine-tuning-based on DistilBERT models. Our results demonstrated evidence of prior probability shifts, covariate shifts, and concept shifts in the data. Applying CL techniques had a statistically significant (α=0.05) positive impact on both backward and forward knowledge transfer, as measured by the F1-score, compared to non-continual approaches. We can argue that the utilization of CL techniques in the context of EMCI is effective in adapting deep learning classifiers to changes in data distributions, thereby maintaining the stability of model performance over time. To our knowledge, this study represents the first exploration of a CL approach using real EMCI data.

Preoperative neurocognitive function as an independent survival prognostic marker in primary glioblastoma.

Background: Aim of the present study is to investigate whether preoperative neurocognitive status is prognostically associated with overall survival (OS) in newly diagnosed glioblastoma (GBM) patients. Methods: Ninety patients with dominant-hemisphere IDH-wild-type GBM were assessed by Mini Mental Status Exam (MMSE), Trail Making Test (TMT) A and B parts, and Control Word Association Test (COWAT) phonemic and semantic subtests. Demographics, Karnofsky Performance Scale, tumor parameters, type of surgery, and adjuvant therapy data were available for patients. Results: According to Cox proportional hazards model the neurocognitive variables of TMT B (P < .01), COWAT semantic subset (P < .05), and the MMSE (P < .01) were found significantly associated with survival prediction. From all other factors, only tumor volume and operation type (debulking vs biopsy) showed a statistical association (P < .05) with survival prediction. Kaplan Meier Long rank test showed statistical significance (P < .01) between unimpaired and impaired groups for TMT B, with median survival for the unimpaired group 26 months and 10 months for the impaired group, for COWAT semantic (P < .01) with median survival 23 months and 12 months, respectively and for MMSE (P < .01) with medial survival 19 and 12 months respectively. Conclusions: Our study demonstrates that neurocognitive status at baseline-prior to treatment-is an independent prognostic factor for OS in wild-type GBM patients, adding another prognostic tool to assist physicians in selecting the best treatment plan.

Extremely missing numerical data in Electronic Health Records for machine learning can be managed through simple imputation methods considering informative missingness: A comparative of solutions in a COVID-19 mortality case study.

BACKGROUND AND OBJECTIVE: Reusing Electronic Health Records (EHRs) for Machine Learning (ML) leads on many occasions to extremely incomplete and sparse tabular datasets, which can hinder the model development processes and limit their performance and generalization. In this study, we aimed to characterize the most effective data imputation techniques and ML models for dealing with highly missing numerical data in EHRs, in the case where only a very limited number of data are complete, as opposed to the usual case of having a reduced number of missing values. METHODS: We used a case study including full blood count laboratory data, demographic and survival data in the context of COVID-19 hospital admissions and evaluated 30 processing pipelines combining imputation methods with ML classifiers. The imputation methods included missing mask, translation and encoding, mean imputation, k-nearest neighbors' imputation, Bayesian ridge regression imputation and generative adversarial imputation networks. The classifiers included k-nearest neighbors, logistic regression, random forest, gradient boosting and deep multilayer perceptron. RESULTS: Our results suggest that in the presence of highly missing data, combining translation and encoding imputation-which considers informative missingness-with tree ensemble classifiers-random forest and gradient boosting-is a sensible choice when aiming to maximize performance, in terms of area under curve. CONCLUSIONS: Based on our findings, we recommend the consideration of this imputer-classifier configuration when constructing models in the presence of extremely incomplete numerical data in EHR.

Vascular differences between IDH-wildtype glioblastoma and astrocytoma IDH-mutant grade 4 at imaging and transcriptomic levels.

Global agreement in central nervous system (CNS) tumor classification is essential for predicting patient prognosis and determining the correct course of treatment, as well as for stratifying patients for clinical trials at international level. The last update by the World Health Organization of CNS tumor classification and grading in 2021 considered, for the first time, IDH-wildtype glioblastoma and astrocytoma IDH-mutant grade 4 as different tumors. Mutations in the genes isocitrate dehydrogenase (IDH) 1 and 2 occur early and, importantly, contribute to gliomagenesis. IDH mutation produces a metabolic reprogramming of tumor cells, thus affecting the processes of hypoxia and vascularity, resulting in a clear advantage for those patients who present with IDH-mutated astrocytomas. Despite the clinical relevance of IDH mutation, current protocols do not include full sequencing for every patient. Alternative biomarkers could be useful and complementary to obtain a more reliable classification. In this sense, magnetic resonance imaging (MRI)-perfusion biomarkers, such as relative cerebral blood volume and flow, could be useful from the moment of presurgery, without incurring additional financial costs or requiring extra effort. The main purpose of this work is to analyze the vascular and hemodynamic differences between IDH-wildtype glioblastoma and IDH-mutant astrocytoma. To achieve this, we evaluate and validate the association between dynamic susceptibility contrast-MRI perfusion biomarkers and IDH mutation status. In addition, to gain a deeper understanding of the vascular differences in astrocytomas depending on the IDH mutation, we analyze the transcriptomic bases of the vascular differences.

Association between preoperative neurocognitive status and IDH1 mutation status in high-grade gliomas.

Background: High-grade glioma (HGG) patients present with variable impairment in neurocognitive function (NCF). Based on that, isocitrate dehydrogenase 1 (IDH1) wild-type HGGs are more aggressive than IDH1 mutant-type ones, we hypothesized that patients with IDH1 wild-type HGG would exhibit more severe NCF deficits than their IDH1 mutant counterparts. Methods: NCF was assessed by Mini Mental Status Exam (MMSE), Trail Making Test (TMT), Digit Span (DS), and Controlled Word Association Test (COWAT) tests in 147 HGG patients preoperatively. Results: Analyses between IDH1 groups revealed a significant difference on MMSE concentration component (p ≤ .01), DS (p ≤ .01), TMTB (p ≤ .01), and COWAT (p ≤ .01) scores, with the IDH1 wild group performing worse than the IDH1 mutant one. Age and tumor volume were inversely correlated with MMSE concentration component (r = -4.78, p < .01), and with MMSE concentration (r = -.401, p < .01), TMTB (r = -.328, p < .01), and COWAT phonemic scores (r = -.599, p < .01), respectively, but only for the IDH1 wild-type group. Analyses between age-matched subsamples of IDH1 groups revealed no age effect on NCF. Tumor grade showed nonsignificance on NCF (p > .05) between the 2 IDH1 mutation subgroups of grade IV tumor patients. On the contrary, grade III group showed a significant difference in TMTB (p < .01) and DS backwards (p < .01) between IDH1 subgroups, with the mutant one outperforming the IDH1 wild one. Conclusions: Our findings indicate that IDH1 wild-type HGG patients present greater NCF impairment, in executive functions particularly, compared to IDH1 mutant ones, suggesting that tumor growth kinetics may play a more profound role than other tumor and demographic parameters in clinical NCF of HGG patients.

User-centred design of a clinical decision support system for palliative care: Insights from healthcare professionals.

Objective: Although clinical decision support systems (CDSS) have many benefits for clinical practice, they also have several barriers to their acceptance by professionals. Our objective in this study was to design and validate The Aleph palliative care (PC) CDSS through a user-centred method, considering the predictions of the artificial intelligence (AI) core, usability and user experience (UX). Methods: We performed two rounds of individual evaluation sessions with potential users. Each session included a model evaluation, a task test and a usability and UX assessment. Results: The machine learning (ML) predictive models outperformed the participants in the three predictive tasks. System Usability Scale (SUS) reported 62.7 ± 14.1 and 65 ± 26.2 on a 100-point rating scale for both rounds, respectively, while User Experience Questionnaire - Short Version (UEQ-S) scores were 1.42 and 1.5 on the -3 to 3 scale. Conclusions: The think-aloud method and including the UX dimension helped us to identify most of the workflow implementation issues. The system has good UX hedonic qualities; participants were interested in the tool and responded positively to it. Performance regarding usability was modest but acceptable.

Complementary frailty and mortality prediction models on older patients as a tool for assessing palliative care needs.

Palliative care (PC) has demonstrated benefits for life-limiting illnesses. Bad survival prognosis and patients' decline are working criteria to guide PC decision-making for older patients. Still, there is not a clear consensus on when to initiate early PC. This work aims to propose machine learning approaches to predict frailty and mortality in older patients in supporting PC decision-making. Predictive models based on Gradient Boosting Machines (GBM) and Deep Neural Networks (DNN) were implemented for binary 1-year mortality classification, survival estimation and 1-year frailty classification. Besides, we tested the similarity between mortality and frailty distributions. The 1-year mortality classifier achieved an Area Under the Curve Receiver Operating Characteristic (AUC ROC) of 0.87 [0.86, 0.87], whereas the mortality regression model achieved an mean absolute error (MAE) of 333.13 [323.10, 342.49] days. Moreover, the 1-year frailty classifier obtained an AUC ROC of 0.89 [0.88, 0.90]. Mortality and frailty criteria were weakly correlated and had different distributions, which can be interpreted as these assessment measurements are complementary for PC decision-making. This study provides new models that can be part of decision-making systems for PC services in older patients after their external validation.

Towards a Computational Approach for the Assessment of Compliance of ALCOA+ Principles in Pharma Industry.

The pharmaceutical industry is a data-intensive environment and a heavily-regulated sector, where exhaustive audits and inspections are performed to ensure the safety of drugs. In this context, processing and evaluating the data generated in the manufacturing lines is a relevant challenge since it requires compliance with pharma regulations. This work combines data integrity metrics and blockchain technology to evaluate the compliance-degree of ALCOA+ principles among different levels of drug manufacturing data. We propose the DIALCOA tool, a software to assess the compliance-degree for each ALCOA+ principle, based on the assessment of data from manufacturing batch reports and its different levels of information.

Discovering Key Topics in Emergency Medical Dispatch from Free Text Dispatcher Observations.

The objective of this work was to discover key topics latent in free text dispatcher observations registered during emergency medical calls. We used a total of 1374931 independent retrospective cases from the Valencian emergency medical dispatch service in Spain, from 2014 to 2019. Text fields were preprocessed to reduce vocabulary size and filter noise, removing accent and punctuation marks, along with uninformative and infrequent words. Key topics were inferred from the multinomial probabilities over words conditioned on each topic from a Latent Dirichlet Allocation model, trained following an online mini-batch variational approach. The optimal number of topics was set analyzing the values of a topic coherence measure, based on the normalized pointwise mutual information, across multiple validation K-folds. Our results support the presence of 15 key topics latent in free text dispatcher observations, related with: ambulance request; chest pain and heart attack; respiratory distress; head falls and blows; fever, chills, vomiting and diarrhea; heart failure; syncope; limb injuries; public service body request; thoracic and abdominal pain; stroke and blood pressure abnormalities; pill intake; diabetes; bleeding; consciousness. The discovery of these topics implies the automatic characterization of a huge volume of complex unstructured data containing relevant information linked to emergency medical call incidents. Hence, results from this work could lead to the update of structured emergency triage algorithms to directly include this latent information in the triage process, resulting in a positive impact in patient wellbeing and health services sustainability.

Subphenotyping of Mexican Patients With COVID-19 at Preadmission To Anticipate Severity Stratification: Age-Sex Unbiased Meta-Clustering Technique.

BACKGROUND: The COVID-19 pandemic has led to an unprecedented global health care challenge for both medical institutions and researchers. Recognizing different COVID-19 subphenotypes-the division of populations of patients into more meaningful subgroups driven by clinical features-and their severity characterization may assist clinicians during the clinical course, the vaccination process, research efforts, the surveillance system, and the allocation of limited resources. OBJECTIVE: We aimed to discover age-sex unbiased COVID-19 patient subphenotypes based on easily available phenotypical data before admission, such as pre-existing comorbidities, lifestyle habits, and demographic features, to study the potential early severity stratification capabilities of the discovered subgroups through characterizing their severity patterns, including prognostic, intensive care unit (ICU), and morbimortality outcomes. METHODS: We used the Mexican Government COVID-19 open data, including 778,692 SARS-CoV-2 population-based patient-level data as of September 2020. We applied a meta-clustering technique that consists of a 2-stage clustering approach combining dimensionality reduction (ie, principal components analysis and multiple correspondence analysis) and hierarchical clustering using the Ward minimum variance method with Euclidean squared distance. RESULTS: In the independent age-sex clustering analyses, 56 clusters supported 11 clinically distinguishable meta-clusters (MCs). MCs 1-3 showed high recovery rates (90.27%-95.22%), including healthy patients of all ages, children with comorbidities and priority in receiving medical resources (ie, higher rates of hospitalization, intubation, and ICU admission) compared with other adult subgroups that have similar conditions, and young obese smokers. MCs 4-5 showed moderate recovery rates (81.30%-82.81%), including patients with hypertension or diabetes of all ages and obese patients with pneumonia, hypertension, and diabetes. MCs 6-11 showed low recovery rates (53.96%-66.94%), including immunosuppressed patients with high comorbidity rates, patients with chronic kidney disease with a poor survival length and probability of recovery, older smokers with chronic obstructive pulmonary disease, older adults with severe diabetes and hypertension, and the oldest obese smokers with chronic obstructive pulmonary disease and mild cardiovascular disease. Group outcomes conformed to the recent literature on dedicated age-sex groups. Mexican states and several types of clinical institutions showed relevant heterogeneity regarding severity, potentially linked to socioeconomic or health inequalities. CONCLUSIONS: The proposed 2-stage cluster analysis methodology produced a discriminative characterization of the sample and explainability over age and sex. These results can potentially help in understanding the clinical patient and their stratification for automated early triage before further tests and laboratory results are available and even in locations where additional tests are not available or to help decide resource allocation among vulnerable subgroups such as to prioritize vaccination or treatments.

Lack of Benefit of Extending Temozolomide Treatment in Patients with High Vascular Glioblastoma with Methylated MGMT.

In this study, we evaluated the benefit on survival of the combination of methylation of O6-methylguanine-DNA methyltransferase (MGMT) promotor gene and moderate vascularity in glioblastoma using a retrospective dataset of 123 patients from a multicenter cohort. MRI processing and calculation of relative cerebral blood volume (rCBV), used to define moderate- and high-vascular groups, were performed with the automatic ONCOhabitats method. We assessed the previously proposed rCBV threshold (10.7) and the new calculated ones (9.1 and 9.8) to analyze the association with survival for different populations according to vascularity and MGMT methylation status. We found that patients included in the moderate-vascular group had longer survival when MGMT is methylated (significant median survival difference of 174 days, p = 0.0129*). However, we did not find significant differences depending on the MGMT methylation status for the high-vascular group (p = 0.9119). In addition, we investigated the combined correlation of MGMT methylation status and rCBV with the prognostic effect of the number of temozolomide cycles, and only significant results were found for the moderate-vascular group. In conclusion, there is a lack of benefit of extending temozolomide treatment for patients with high vascular glioblastomas, even presenting MGMT methylation. Preliminary results suggest that patients with moderate vascularity and methylated MGMT glioblastomas would benefit more from prolonged adjuvant chemotherapy.

Quality of Hospital Electronic Health Record (EHR) Data Based on the International Consortium for Health Outcomes Measurement (ICHOM) in Heart Failure: Pilot Data Quality Assessment Study.

BACKGROUND: There is increasing recognition that health care providers need to focus attention, and be judged against, the impact they have on the health outcomes experienced by patients. The measurement of health outcomes as a routine part of clinical documentation is probably the only scalable way of collecting outcomes evidence, since secondary data collection is expensive and error-prone. However, there is uncertainty about whether routinely collected clinical data within electronic health record (EHR) systems includes the data most relevant to measuring and comparing outcomes and if those items are collected to a good enough data quality to be relied upon for outcomes assessment, since several studies have pointed out significant issues regarding EHR data availability and quality. OBJECTIVE: In this paper, we first describe a practical approach to data quality assessment of health outcomes, based on a literature review of existing frameworks for quality assessment of health data and multistakeholder consultation. Adopting this approach, we performed a pilot study on a subset of 21 International Consortium for Health Outcomes Measurement (ICHOM) outcomes data items from patients with congestive heart failure. METHODS: All available registries compatible with the diagnosis of heart failure within an EHR data repository of a general hospital (142,345 visits and 12,503 patients) were extracted and mapped to the ICHOM format. We focused our pilot assessment on 5 commonly used data quality dimensions: completeness, correctness, consistency, uniqueness, and temporal stability. RESULTS: We found high scores (>95%) for the consistency, completeness, and uniqueness dimensions. Temporal stability analyses showed some changes over time in the reported use of medication to treat heart failure, as well as in the recording of past medical conditions. Finally, the investigation of data correctness suggested several issues concerning the characterization of missing data values. Many of these issues appear to be introduced while mapping the IMASIS-2 relational database contents to the ICHOM format, as the latter requires a level of detail that is not explicitly available in the coded data of an EHR. CONCLUSIONS: Overall, results of this pilot study revealed good data quality for the subset of heart failure outcomes collected at the Hospital del Mar. Nevertheless, some important data errors were identified that were caused by fundamentally different data collection practices in routine clinical care versus research, for which the ICHOM standard set was originally developed. To truly examine to what extent hospitals today are able to routinely collect the evidence of their success in achieving good health outcomes, future research would benefit from performing more extensive data quality assessments, including all data items from the ICHOM standards set and across multiple hospitals.

Deep ensemble multitask classification of emergency medical call incidents combining multimodal data improves emergency medical dispatch.

The objective of this work was to develop a predictive model to aid non-clinical dispatchers to classify emergency medical call incidents by their life-threatening level (yes/no), admissible response delay (undelayable, minutes, hours, days) and emergency system jurisdiction (emergency system/primary care) in real time. We used a total of 1 244 624 independent incidents from the Valencian emergency medical dispatch service in Spain, compiled in retrospective from 2009 to 2012, including clinical features, demographics, circumstantial factors and free text dispatcher observations. Based on them, we designed and developed DeepEMC2, a deep ensemble multitask model integrating four subnetworks: three specialized to context, clinical and text data, respectively, and another to ensemble the former. The four subnetworks are composed in turn by multi-layer perceptron modules, bidirectional long short-term memory units and a bidirectional encoding representations from transformers module. DeepEMC2 showed a macro F1-score of 0.759 in life-threatening classification, 0.576 in admissible response delay and 0.757 in emergency system jurisdiction. These results show a substantial performance increase of 12.5 %, 17.5 % and 5.1 %, respectively, with respect to the current in-house triage protocol of the Valencian emergency medical dispatch service. Besides, DeepEMC2 significantly outperformed a set of baseline machine learning models, including naive bayes, logistic regression, random forest and gradient boosting (α = 0.05). Hence, DeepEMC2 is able to: 1) capture information present in emergency medical calls not considered by the existing triage protocol, and 2) model complex data dependencies not feasible by the tested baseline models. Likewise, our results suggest that most of this unconsidered information is present in the free text dispatcher observations. To our knowledge, this study describes the first deep learning model undertaking emergency medical call incidents classification. Its adoption in medical dispatch centers would potentially improve emergency dispatch processes, resulting in a positive impact in patient wellbeing and health services sustainability.

Predicting morbidity by local similarities in multi-scale patient trajectories.

Patient Trajectories (PTs) are a method of representing the temporal evolution of patients. They can include information from different sources and be used in socio-medical or clinical domains. PTs have generally been used to generate and study the most common trajectories in, for instance, the development of a disease. On the other hand, healthcare predictive models generally rely on static snapshots of patient information. Only a few works about prediction in healthcare have been found that use PTs, and therefore benefit from their temporal dimension. All of them, however, have used PTs created from single-source information. Therefore, the use of longitudinal multi-scale data to build PTs and use them to obtain predictions about health conditions is yet to be explored. Our hypothesis is that local similarities on small chunks of PTs can identify similar patients concerning their future morbidities. The objectives of this work are (1) to develop a methodology to identify local similarities between PTs before the occurrence of morbidities to predict these on new query individuals; and (2) to validate this methodology on risk prediction of cardiovascular diseases (CVD) occurrence in patients with diabetes. We have proposed a novel formal definition of PTs based on sequences of longitudinal multi-scale data. Moreover, a dynamic programming methodology to identify local alignments on PTs for predicting future morbidities is proposed. Both the proposed methodology for PT definition and the alignment algorithm are generic to be applied on any clinical domain. We validated this solution for predicting CVD in patients with diabetes and we achieved a precision of 0.33, a recall of 0.72 and a specificity of 0.38. Therefore, the proposed solution in the diabetes use case can result of utmost utility to secondary screening.

How the Wakamola chatbot studied a university community's lifestyle during the COVID-19 confinement.

This work aimed to study the effect of confinement on weight and lifestyle using the Wakamola chatbot to collect data from 739 adults divided into two groups (341 case-control, 398 confinement). Nutrition score (0-100 scale) improved for men (medians 81.77-82.29, p < 0.05), with no difference for women (medians 82.29 in both cases). Both genders reduced the consumption of sweetmeats and sugared drinks (p < 0.01); men increased their consumption of vegetables, salad, and legumes (p < 0.01). Both genders reduced their physical activity score (men 100-40.14, p < 0.01, women 80.42-36.12, p < 0.01). Women sat less hours/week, men's medians 28.81-28.27, women's medians 35.97-23.33, p = 0.03. Both genders slept longer (hours/day), men 7-7.5, women 7-8 (p < 0.01) (medians). Their overall health score was significantly reduced (men 85.06-74.05, p < 0.01, women 84.47-72.42, p < 0.01), with no significant weight difference in either gender. Wakamola helped to contact participants and confirm changes in their lifestyle during confinement.

A User-Centered Chatbot (Wakamola) to Collect Linked Data in Population Networks to Support Studies of Overweight and Obesity Causes: Design and Pilot Study.

BACKGROUND: Obesity and overweight are a serious health problem worldwide with multiple and connected causes. Simultaneously, chatbots are becoming increasingly popular as a way to interact with users in mobile health apps. OBJECTIVE: This study reports the user-centered design and feasibility study of a chatbot to collect linked data to support the study of individual and social overweight and obesity causes in populations. METHODS: We first studied the users' needs and gathered users' graphical preferences through an open survey on 52 wireframes designed by 150 design students; it also included questions about sociodemographics, diet and activity habits, the need for overweight and obesity apps, and desired functionality. We also interviewed an expert panel. We then designed and developed a chatbot. Finally, we conducted a pilot study to test feasibility. RESULTS: We collected 452 answers to the survey and interviewed 4 specialists. Based on this research, we developed a Telegram chatbot named Wakamola structured in six sections: personal, diet, physical activity, social network, user's status score, and project information. We defined a user's status score as a normalized sum (0-100) of scores about diet (frequency of eating 50 foods), physical activity, BMI, and social network. We performed a pilot to evaluate the chatbot implementation among 85 healthy volunteers. Of 74 participants who completed all sections, we found 8 underweight people (11%), 5 overweight people (7%), and no obesity cases. The mean BMI was 21.4 kg/m2 (normal weight). The most consumed foods were olive oil, milk and derivatives, cereals, vegetables, and fruits. People walked 10 minutes on 5.8 days per week, slept 7.02 hours per day, and were sitting 30.57 hours per week. Moreover, we were able to create a social network with 74 users, 178 relations, and 12 communities. CONCLUSIONS: The Telegram chatbot Wakamola is a feasible tool to collect data from a population about sociodemographics, diet patterns, physical activity, BMI, and specific diseases. Besides, the chatbot allows the connection of users in a social network to study overweight and obesity causes from both individual and social perspectives.

Design of 1-year mortality forecast at hospital admission: A machine learning approach.

Palliative care is referred to a set of programs for patients that suffer life-limiting illnesses. These programs aim to maximize the quality of life (QoL) for the last stage of life. They are currently based on clinical evaluation of the risk of 1-year mortality. The main aim of this work is to develop and validate machine-learning-based models to predict the exitus of a patient within the next year using data gathered at hospital admission. Five machine-learning techniques were applied using a retrospective dataset. The evaluation was performed with five metrics computed by a resampling strategy: Accuracy, the area under the ROC curve, Specificity, Sensitivity, and the Balanced Error Rate. All models reported an AUC ROC from 0.857 to 0.91. Specifically, Gradient Boosting Classifier was the best model, producing an AUC ROC of 0.91, a sensitivity of 0.858, a specificity of 0.808, and a BER of 0.1687. Information from standard procedures at hospital admission combined with machine learning techniques produced models with competitive discriminative power. Our models reach the best results reported in the state of the art. These results demonstrate that they can be used as an accurate data-driven palliative care criteria inclusion.

Potential limitations in COVID-19 machine learning due to data source variability: A case study in the nCov2019 dataset.

OBJECTIVE: The lack of representative coronavirus disease 2019 (COVID-19) data is a bottleneck for reliable and generalizable machine learning. Data sharing is insufficient without data quality, in which source variability plays an important role. We showcase and discuss potential biases from data source variability for COVID-19 machine learning. MATERIALS AND METHODS: We used the publicly available nCov2019 dataset, including patient-level data from several countries. We aimed to the discovery and classification of severity subgroups using symptoms and comorbidities. RESULTS: Cases from the 2 countries with the highest prevalence were divided into separate subgroups with distinct severity manifestations. This variability can reduce the representativeness of training data with respect the model target populations and increase model complexity at risk of overfitting. CONCLUSIONS: Data source variability is a potential contributor to bias in distributed research networks. We call for systematic assessment and reporting of data source variability and data quality in COVID-19 data sharing, as key information for reliable and generalizable machine learning.

MGMT methylation may benefit overall survival in patients with moderately vascularized glioblastomas.

OBJECTIVES: To assess the combined role of tumor vascularity, estimated from perfusion MRI, and MGMT methylation status on overall survival (OS) in patients with glioblastoma. METHODS: A multicentric international dataset including 96 patients from NCT03439332 clinical study were used to study the prognostic relationships between MGMT and perfusion markers. Relative cerebral blood volume (rCBV) in the most vascularized tumor regions was automatically obtained from preoperative MRIs using ONCOhabitats online analysis service. Cox survival regression models and stratification strategies were conducted to define a subpopulation that is particularly favored by MGMT methylation in terms of OS. RESULTS: rCBV distributions did not differ significantly (p > 0.05) in the methylated and the non-methylated subpopulations. In patients with moderately vascularized tumors (rCBV < 10.73), MGMT methylation was a positive predictive factor for OS (HR = 2.73, p = 0.003, AUC = 0.70). In patients with highly vascularized tumors (rCBV > 10.73), however, there was no significant effect of MGMT methylation (HR = 1.72, p = 0.10, AUC = 0.56). CONCLUSIONS: Our results indicate the existence of complementary prognostic information provided by MGMT methylation and rCBV. Perfusion markers could identify a subpopulation of patients who will benefit the most from MGMT methylation. Not considering this information may lead to bias in the interpretation of clinical studies. KEY POINTS: • MRI perfusion provides complementary prognostic information to MGMT methylation. • MGMT methylation improves prognosis in glioblastoma patients with moderate vascular profile. • Failure to consider these relations may lead to bias in the interpretation of clinical studies.