We present a model for predicting electrocardiogram (ECG) abnormalities in shortduration 12-lead ECG signals which outperformed medical doctors on the 4th year of their cardiology residency. Such exams can provide a full evaluation of heart activity and have not been studied in previous end-to-end machine learning papers. Using the database of a large telehealth network, we built a novel dataset with more than 2 million ECG tracings, orders of magnitude larger than those used in previous studies. Moreover, our dataset is more realistic, as it consist of 12-lead ECGs recorded during standard in-clinics exams. Using this data, we trained a residual neural network with 9 convolutional layers to map 7 to 10 second ECG signals to 6 classes of ECG abnormalities. Future work should extend these results to cover a large range of ECG abnormalities, which could improve the accessibility of this diagnostic tool and avoid wrong diagnosis from medical doctors. IntroductionCardiovascular diseases are the leading cause of death worldwide [1] and the electrocardiogram (ECG) is a major diagnostic tool for this group of diseases. As ECGs transitioned from analogue to digital, automated computer analysis of standard 12-lead electrocardiograms gained importance in the process of medical diagnosis [2]. However, limited performance of classical algorithms [3,4] precludes its usage as a standalone diagnostic tool and relegates it to an ancillary role [5].End-to-end deep learning has recently achieved striking success in task such as image classification [6] and speech recognition [7], and there are great expectations about how this technology may improve health care and clinical practice [8][9][10]. So far, the most successful applications used a supervised learning setup to automate diagnosis from exams. Algorithms have achieved better performance than a human specialist on their routine workflow in diagnosing breast cancer [11] and detecting certain eye conditions from eye scans [12]. While efficient, training deep neural networks using supervised learning algorithms introduces the need for large quantities of labeled data which, for medical applications, introduce several challenges, including those related to confidentiality and security of personal health information [13].Standard, short-duration 12-lead ECG is the most commonly used complementary exam for the evaluation of the heart, being employed across all clinical settings: from the primary care centers to the intensive care units. While tracing cardiac monitors and long-term monitoring, as the Holter exam, provides information mostly about cardiac rhythm and repolarization, 12-lead ECG can provide a full evaluation of heart, including arrhythmias, conduction disturbances, acute coronary syndromes, cardiac chamber hypertrophy and enlargement and even the effects of drugs and electrolyte disturbances.Machine Learning for Health (ML4H) Workshop at NeurIPS 2018.

The electrocardiogram (ECG) is the most commonly used exam for the evaluation of cardiovascular diseases. Here we propose that the age predicted by artificial intelligence (AI) from the raw ECG (ECG-age) can be a measure of cardiovascular health. A deep neural network is trained to predict a patient’s age from the 12-lead ECG in the CODE study cohort (n = 1,558,415 patients). On a 15% hold-out split, patients with ECG-age more than 8 years greater than the chronological age have a higher mortality rate (hazard ratio (HR) 1.79, p < 0.001), whereas those with ECG-age more than 8 years smaller, have a lower mortality rate (HR 0.78, p < 0.001). Similar results are obtained in the external cohorts ELSA-Brasil (n = 14,236) and SaMi-Trop (n = 1,631). Moreover, even for apparent normal ECGs, the predicted ECG-age gap from the chronological age remains a statistically significant risk predictor. These results show that the AI-enabled analysis of the ECG can add prognostic information.

Inselbergs are rocky environments that support a unique flora distinct from that of the surrounding area. The objectives of this work were to conduct a floristic inventory of an inselberg in the semi-arid region of Northeast Brazil, and to investigate the relationship between its flora and the flora of the surrounding area. The following questions were addressed: is the species richness comparable to other inselbergs in northeastern Brazil; is the floristic composition of the inselberg more similar to other inselbergs or to the surrounding Caatinga vegetation; and do the similarities in the floristic composition of inselbergs depend on the distance between them? This work documents 201 species in 62 families. Cyperaceae (28 spp.), Euphorbiaceae (19), Poaceae (15), Orchidaceae (11) and Bromeliaceae (9) are the most species-rich families. On the inselberg the plants are distributed in islands found on exposed rock, in fissures and in depressions in the rock. Variations in species richness in the region were assessed by comparison of floristic inventories conducted in other inselbergs of the semi-arid region with those of this study. The flora of the inselberg under investigation is more similar to the flora of other nearby inselbergs than to the vegetation of the surrounding semi-arid region.

Digital electrocardiographs are now widely available and a large number of digital electrocardiograms (ECGs) have been recorded and stored. The present study describes the development and clinical applications of a large database of such digital ECGs, namely the CODE (Clinical Outcomes in Digital Electrocardiology) study. ECGs obtained by the Telehealth Network of Minas Gerais, Brazil, from 2010-17, were organized in a structured database. A hierarchical free-text machine learning algorithm recognized specific ECG diagnoses from cardiologist reports. The Glasgow ECG Analysis Program provided Minnesota Codes and automatic diagnostic statements. The presence of a specific ECG abnormality was considered when both automatic and medical diagnosis were concordant; cases of discordance were decided using heuristisc rules and manual review. The ECG database was linked to the national mortality information system using probabilistic linkage methods. From 2,470,424 ECGs, 1,773,689 patients were identified. After excluding the ECGs with technical problems and patients <16 years-old, 1,558,415 patients were studied. High performance measures were obtained using an end-to-end deep neural network trained to detect 6 types of ECG abnormalities, with F1 scores >80% and specificity >99% in an independent test dataset. We also evaluated the risk of mortality associated with the presence of atrial fibrillation (AF), which showed that AF was a strong predictor of cardiovascular mortality and mortality for all causes, with increased risk in women. In conclusion, a large database that comprises all ECGs performed by a large telehealth network can be useful for further developments in the field of digital electrocardiography, clinical cardiology and cardiovascular epidemiology.

BackgroundKnowledge of the normal limits of the electrocardiogram (ECG) is mandatory for establishing which patients have abnormal ECGs. No studies have assessed the reference standards for a Latin American population. Our aim was to establish the normal ranges of the ECG for pediatric and adult Brazilian primary care patients.MethodsThis retrospective observational study assessed all the consecutive 12-lead digital electrocardiograms of primary care patients at least 1 year old in Minas Gerais state, Brazil, recorded between 2010 and 2015. ECGs were excluded if there were technical problems, selected abnormalities were present or patients with selected self-declared comorbidities or on drug therapy. Only the first ECG from patients with multiple ECGs was accepted. The University of Glasgow ECG analysis program was used to automatically interpret the ECGs. For each variable, the 1st, 2nd, 50th, 98th and 99th percentiles were determined and results were compared to selected studies.ResultsA total of 1,493,905 ECGs were recorded. 1,007,891 were excluded and 486.014 were analyzed. This large study provided normal values for heart rate, P, QRS and T frontal axis, P and QRS overall duration, PR and QT overall intervals and QTc corrected by Hodges, Bazett, Fridericia and Framingham formulae. Overall, the results were similar to those from other studies performed in different populations but there were differences in extreme ages and specific measurements.ConclusionsThis study has provided reference values for Latinos of both sexes older than 1 year. Our results are comparable to studies performed in different populations.

Introduction Mobile-technology-based interventions are promising strategies for promoting behavioural change in obese patients. The aims of this study were to evaluate the feasibility of implementing a text message intervention, and to assess the effects of the intervention on body mass index (BMI) and self-reported behavioural change. Methods TELEFIT was a three-phase feasibility study comprising the following stages: (a) the development of text messages; (b) testing; and (c) a quasi-experimental pilot study in which patients who were engaged in obesity/overweight educational groups in public primary care centres in Belo Horizonte, Brazil, were recruited. A bank of text messages was drafted and reviewed by an expert panel, text message delivery software was developed and tested, and a pilot study assessed patients before and after receiving the intervention using validated questionnaires and body measures. The data were analysed using the Wilcoxon test. Results A total of 46 patients completed the follow-up; 93.5% were women and the median age was 42 years (interquartile range (IQR) 34-52 years). At four months, participants had a significant reduction in BMI (median 31.3 (IQR 28.2-34.6) vs. 29.9 (IQR 27.2-34.6) kg/m, p < 0.001), systolic (median 125 (IQR 120-132) vs. 120 (IQR 110-130) mmHg, p = 0.013) and diastolic blood pressure (median 80 (IQR 70-100) vs. 80 (IQR 70-80) mmHg, p = 0.006), when compared to baseline. All patients reported to be satisfied and willing to continue receiving the intervention, and 93.3% felt that the intervention helped them change their behaviours. Discussion This study has shown that a text message intervention to promote behavioural change and weight loss was feasible and effective in a short-term period. Participants were satisfied and willing to continue receiving the SMS messages.

-(Floristic diversity of two crystalline rocky outcrops in the Brazilian northeast semi-arid region). Floristic composition and structure of vegetation were studied in two rocky outcrop areas in the semi-arid region of northeastern Brazil. From April 2007 to September 2008, 18 monthly field trips were carried out. Vascular plants were randomly collected throughout the outcrop areas. For structural analysis, 30 plots of 1 × 1 m were set in the vegetation islands. The checklist presented combines 211 species (69 families and 168 genera), although only 56 species were collected in the plots. Fabaceae (18 spp.; 8.5%), Asteraceae (17 spp.; 8%), Orchidaceae (13 spp.; 6.1%), Euphorbiaceae (13 spp.; 6.1%), Bromeliaceae (10 spp.; 4.7%), and Poaceae (eight spp.; 3.8%) are the richest families. Overall, 1,792 shrub and herbaceous specimens were counted in the plots. The Shannon-Wiener (H) diversity index values were 2.572 and 2.547 nats individual -1 . The species that presented the highest absolute abundance values (number of plants) had low frequencies in the plots and vice-versa. The biological spectrum had a high proportion of phanerophytes and therophytes, followed by cryptophytes, chamaephytes, and hemicryptophytes. The studied flora shares floristic components similar to other rocky outcrop areas of the semi-arid region in northeastern Brazil, including in relation to dominant groups in the vegetation structure.Key words -Caatinga, inselbergs, life-forms, saxicolous plants RESUMO -(Diversidade florística de dois afloramentos rochosos cristalinos no semi-árido, nordeste do Brasil).Composição florística e estrutura da vegetação foram estudadas em dois afloramentos rochosos localizados no semi-árido do nordeste do Brasil. Foram realizadas 18 excursões mensais, de abril de 2007 a setembro de 2008. Plantas vasculares foram coletadas aleatoriamente, no afloramento como um todo. Para análise estrutural foram plotadas 30 parcelas de 1 × 1 m nas ilhas de vegetação. Foram encontradas 211 espécies (69 famílias e 168 gêneros), entretanto somente 56 espécies foram coletadas nas parcelas. As famílias com maior número de espécies foram Fabaceae (18 spp.; 9%), Asteraceae (17 spp.; 8,5%), Orchidaceae (13 spp.; 6,5%), Euphorbiaceae (13 spp.; 6,5%), Bromeliaceae (10 spp.; 5%) e Poaceae (oito spp.; 4%). Ao todo, foram contabilizados, nas parcelas, 1.792 indivíduos herbáceos e arbustivos. Os valores do índice de diversidade de Shannon-Wiener (H) foram de 2,572 e 2,547 nats ind.-1 . As espécies que apresentaram maiores densidade apresentaram baixa freqüência nas parcelas e vice-versa. O espectro biológico apresentou alta proporção de fanerófitos e terófitos, seguidos por criptófitos, caméfitos e hemi-criptófitos. A flora estudada compartilha conjunto florístico semelhante a outros afloramentos rochosos do Nordeste do Brasil, inclusive em termos de grupos dominantes na estrutura da vegetação.Palavras-chave -Caatinga, formas de vida, inselbergues, plantas saxícolas 1.

The electrocardiogram (ECG) is the most commonly used exam for the screening and evaluation of cardiovascular diseases. Here we propose that the age predicted by artificial intelligence (AI) from the raw ECG tracing (ECG-age) can be a measure of cardiovascular health and provide prognostic information. A deep convolutional neural network was trained to predict a patient's age from the 12-lead ECG using data from patients that underwent an ECG from 2010 to 2017 - the CODE study cohort (n=1,558,415 patients). On the 15% hold-out CODE test split, patients with ECG-age more than 8 years greater than chronological age had a higher mortality rate (hazard ratio (HR) 1.79, p<0.001) in a mean follow-up of 3.67 years, whereas those with ECG-age more than 8 years less than chronological age had a lower mortality rate (HR 0.78, p<0.001). Similar results were obtained in the external cohorts ELSA-Brasil (n=14,236) and SaMi-Trop (n=1,631). The ability to predict mortality from the ECG predicted age remains even when we adjust the model for cardiovascular risk factors. Moreover, even for apparent normal ECGs, having a predicted ECG-age 8 or more years greater than chronological age remained a statistically significant predictor of risk (HR 1.53, p<0.001 in CODE 15% test split). These results show that AI-enabled analysis of the ECG can add prognostic information to the interpretation of the 12-lead ECGs.