RESUMEN
PURPOSE: This work aims to assess standard evaluation practices used by the research community for evaluating medical imaging classifiers, with a specific focus on the implications of class imbalance. The analysis is performed on chest X-rays as a case study and encompasses a comprehensive model performance definition, considering both discriminative capabilities and model calibration. MATERIALS AND METHODS: We conduct a concise literature review to examine prevailing scientific practices used when evaluating X-ray classifiers. Then, we perform a systematic experiment on two major chest X-ray datasets to showcase a didactic example of the behavior of several performance metrics under different class ratios and highlight how widely adopted metrics can conceal performance in the minority class. RESULTS: Our literature study confirms that: (1) even when dealing with highly imbalanced datasets, the community tends to use metrics that are dominated by the majority class; and (2) it is still uncommon to include calibration studies for chest X-ray classifiers, albeit its importance in the context of healthcare. Moreover, our systematic experiments confirm that current evaluation practices may not reflect model performance in real clinical scenarios and suggest complementary metrics to better reflect the performance of the system in such scenarios. CONCLUSION: Our analysis underscores the need for enhanced evaluation practices, particularly in the context of class-imbalanced chest X-ray classifiers. We recommend the inclusion of complementary metrics such as the area under the precision-recall curve (AUC-PR), adjusted AUC-PR, and balanced Brier score, to offer a more accurate depiction of system performance in real clinical scenarios, considering metrics that reflect both, discrimination and calibration performance. CLINICAL RELEVANCE STATEMENT: This study underscores the critical need for refined evaluation metrics in medical imaging classifiers, emphasizing that prevalent metrics may mask poor performance in minority classes, potentially impacting clinical diagnoses and healthcare outcomes. KEY POINTS: Common scientific practices in papers dealing with X-ray computer-assisted diagnosis (CAD) systems may be misleading. We highlight limitations in reporting of evaluation metrics for X-ray CAD systems in highly imbalanced scenarios. We propose adopting alternative metrics based on experimental evaluation on large-scale datasets.
RESUMEN
Voice disorders, such as dysphonia, are common among the general population. These pathologies often remain untreated until they reach a high level of severity. Assisting the detection of voice disorders could facilitate early diagnosis and subsequent treatment. In this study, we address the practical aspects of automatic voice disorders detection (AVDD). In real-world scenarios, data annotated for voice disorders is usually scarce due to various challenges involved in the collection and annotation of such data. However, some relatively large datasets are available for a reduced number of domains. In this context, we propose the use of a combination of out-of-domain and in-domain data for training a deep neural network-based AVDD system, and offer guidance on the minimum amount of in-domain data required to achieve acceptable performance. Further, we propose the use of a cost-based metric, the normalized expected cost (EC), to evaluate performance of AVDD systems in a way that closely reflects the needs of the application. As an added benefit, optimal decisions for the EC can be made in a principled way given by Bayes decision theory. Finally, we argue that for medical applications like AVDD, the categorical decisions need to be accompanied by interpretable scores that reflect the confidence of the system. Even very accurate models often produce scores that are not suited for interpretation. Here, we show that such models can be easily improved by adding a calibration stage-trained with just a few minutes of in-domain data. The outputs of the resulting calibrated system can then better support practitioners in their decision-making process.
RESUMEN
Validation metrics are key for tracking scientific progress and bridging the current chasm between artificial intelligence research and its translation into practice. However, increasing evidence shows that, particularly in image analysis, metrics are often chosen inadequately. Although taking into account the individual strengths, weaknesses and limitations of validation metrics is a critical prerequisite to making educated choices, the relevant knowledge is currently scattered and poorly accessible to individual researchers. Based on a multistage Delphi process conducted by a multidisciplinary expert consortium as well as extensive community feedback, the present work provides a reliable and comprehensive common point of access to information on pitfalls related to validation metrics in image analysis. Although focused on biomedical image analysis, the addressed pitfalls generalize across application domains and are categorized according to a newly created, domain-agnostic taxonomy. The work serves to enhance global comprehension of a key topic in image analysis validation.
Asunto(s)
Inteligencia ArtificialRESUMEN
Increasing evidence shows that flaws in machine learning (ML) algorithm validation are an underestimated global problem. In biomedical image analysis, chosen performance metrics often do not reflect the domain interest, and thus fail to adequately measure scientific progress and hinder translation of ML techniques into practice. To overcome this, we created Metrics Reloaded, a comprehensive framework guiding researchers in the problem-aware selection of metrics. Developed by a large international consortium in a multistage Delphi process, it is based on the novel concept of a problem fingerprint-a structured representation of the given problem that captures all aspects that are relevant for metric selection, from the domain interest to the properties of the target structure(s), dataset and algorithm output. On the basis of the problem fingerprint, users are guided through the process of choosing and applying appropriate validation metrics while being made aware of potential pitfalls. Metrics Reloaded targets image analysis problems that can be interpreted as classification tasks at image, object or pixel level, namely image-level classification, object detection, semantic segmentation and instance segmentation tasks. To improve the user experience, we implemented the framework in the Metrics Reloaded online tool. Following the convergence of ML methodology across application domains, Metrics Reloaded fosters the convergence of validation methodology. Its applicability is demonstrated for various biomedical use cases.
Asunto(s)
Algoritmos , Procesamiento de Imagen Asistido por Computador , Aprendizaje Automático , SemánticaRESUMEN
Validation metrics are key for the reliable tracking of scientific progress and for bridging the current chasm between artificial intelligence (AI) research and its translation into practice. However, increasing evidence shows that particularly in image analysis, metrics are often chosen inadequately in relation to the underlying research problem. This could be attributed to a lack of accessibility of metric-related knowledge: While taking into account the individual strengths, weaknesses, and limitations of validation metrics is a critical prerequisite to making educated choices, the relevant knowledge is currently scattered and poorly accessible to individual researchers. Based on a multi-stage Delphi process conducted by a multidisciplinary expert consortium as well as extensive community feedback, the present work provides the first reliable and comprehensive common point of access to information on pitfalls related to validation metrics in image analysis. Focusing on biomedical image analysis but with the potential of transfer to other fields, the addressed pitfalls generalize across application domains and are categorized according to a newly created, domain-agnostic taxonomy. To facilitate comprehension, illustrations and specific examples accompany each pitfall. As a structured body of information accessible to researchers of all levels of expertise, this work enhances global comprehension of a key topic in image analysis validation.
RESUMEN
ABSTRACT: Objective: To compare the effects of the two techniques (minimally invasive transforaminal inter somatic lumbar fusion [MIS-TLIF] and open transforaminal lumbar interbody fusion [TLIF]) in the treatment of lumbar degenerative disease. Methods: This is a retrospective cohort study. The outcomes investigated were: intensity of low back pain, functional disability of the lumbar spine, discharge time, return to work, lumbar lordosis angle, cost of individual sources due to the period of work-related absenteeism, and societal perspective costs in the treatment of low-grade lumbar degenerative disease. The data was obtained through the analysis of data contained in the electronic medical records of 100 patients who underwent one of the two surgical techniques from January 2019 to May 2021 in a High Complexity Orthopedic Surgery. The outcomes investigated were set 12 months postoperatively. Results: No statistical differences were observed in terms of sex, age, employment, and diagnosis grade between groups. MIS-TLIF was associated with significant improvement in the intensity of low back pain, functional disability of the lumbar spine, discharge time, return to work, cost of individual sources due to the period of work-related absenteeism, and societal perspective costs. The variation in the lumbar lordosis angle of the MIS-TLIF group was smaller when compared to TLIF. Conclusion: Considering that MIS-TLIF was achieved with satisfactory short-term improvements, it may be used as an alternative strategy to TLIF to promote clinical and economical improvements in treating lumbar degenerative disease. Level of Evidence III; Comparative Retrospective Study.
RESUMO: Objetivo: Comparar os efeitos das duas técnicas (fusão intersomática lombar transforaminal minimamente invasiva [MIS-TLIF] e fusão intersomática lombar transforaminal aberta [TLIF]) no tratamento da doença degenerativa lombar. Métodos: Trata-se de um estudo de coorte retrospectivo. Os desfechos investigados foram: intensidade da dor lombar, incapacidade funcional da coluna lombar, tempo de alta, retorno ao trabalho, ângulo de lordose lombar, costa de fontes individuais devido ao período de absenteísmo relacionado ao trabalho e custos da perspectiva social no tratamento de doença degenerativa lombar de baixo grau. Os dados foram obtidos por meio da análise dos dados contidos nos prontuários eletrônicos de 100 pacientes submetidos a uma das duas técnicas cirúrgicas no período de janeiro de 2019 a maio de 2021 em uma Cirurgia Ortopédica de Alta Complexidade. Os resultados investigados foram definidos 12 meses após a cirurgia. Resultados: Não foram observadas diferenças estatísticas quanto ao sexo, idade, vínculo empregatício, grau de diagnóstico entre os grupos. O MIS-TLIF foi associado a melhora significativa na intensidade da dor lombar, incapacidade funcional da coluna lombar, tempo de alta, retorno ao trabalho, custo de fontes individuais devido ao período de absenteísmo relacionado ao trabalho e custos de perspectiva social. A variação do ângulo da lordose lombar do grupo MIS-TLIF foi menor quando comparado ao TLIF. Conclusão: Considerando que o MIS-TLIF foi alcançado com melhoras satisfatórias em curto prazo, pode ser usado como uma estratégia alternativa ao TLIF para promover melhorias clínicas e econômicas no tratamento da doença degenerativa lombar. Nível de Evidência III; Estudo Retrospectivo Comparativo.
RESUMEN: Objetivo: Comparar los efectos de dos técnicas (fusión lumbar intersomática transforaminal mínimamente invasiva [MIS-TLIF] y fusión intersomática lumbar transforaminal abierta [TLIF]) en el tratamiento de la enfermedad degenerativa lumbar. Métodos: Este es un estudio de cohorte retrospectivo. Los desenlaces investigados fueron: intensidad del dolor lumbar, incapacidad funcional de la columna lumbar, tiempo de alta, regreso al trabajo, ángulo de lordosis lumbar, costa de fuentes individuales debido al período de ausentismo relacionado con el trabajo y costos de perspectiva social en el tratamiento de enfermedad degenerativa lumbar de bajo grado. Los datos se obtuvieron a través del análisis de los datos contenidos en las historias clínicas electrónicas de 100 pacientes que se sometieron a alguna de las dos técnicas quirúrgicas durante el período de enero de 2019 a mayo de 2021 en una Cirugía Ortopédica de Alta Complejidad. Los resultados investigados se establecieron 12 meses después de la operación. Resultados: No se observaron diferencias estadísticas en cuanto a sexo, edad, ocupación, grado de diagnóstico entre grupos. MIS-TLIF se asoció con una mejora significativa en la intensidad del dolor lumbar, la discapacidad funcional de la columna lumbar, el tiempo de alta, el regreso al trabajo, la costa de fuentes individuales debido al período de ausentismo relacionado con el trabajo y los costos de perspectiva social. La variación en el ángulo de lordosis lumbar del grupo MIS-TLIF fue menor en comparación con TLIF. Conclusión: considerando que MIS-TLIF se logró con mejoras satisfactorias a corto plazo, puede usarse como una estrategia alternativa a TLIF para promover mejoras clínicas y económicas en el tratamiento de la enfermedad degenerativa lumbar. Nivel de Evidencia III; Estudio Retrospectivo Comparativo.
Asunto(s)
Humanos , OrtopediaRESUMEN
BACKGROUND: Few studies have described the epidemiology and clinical behavior of inflammatory bowel disease (IBD) in South America. The aim of this study was to report on the prevalence, phenotype, and treatment of patients with IBD diagnosis in Capital Department of the Province of Córdoba, Argentina. METHODS: Data from adult patients (≥ 18 years-old) with IBD diagnosis that attended 12 public or private centers between 05/2014 and 05/2019 were included in a common registry. RESULTS: A total of 655 patients were included (females: 53.4%). The ratio of ulcerative colitis (UC) (nâ¯=â¯561) to Crohn's disease (CD) (nâ¯=â¯88) was 6.38, with age-adjusted IBD prevalence being 70.1 (95% confidence interval 70.08-70.12) cases/100,000 habitants. Extraintestinal manifestations were diagnosed in 22.8% of patients, and left-side colitis (46%) was the most frequent extension in UC patients. In CD patients, colonic involvement (55.7%) and non-stricturing/non-penetrating behavior (74%) were the most frequent presentations. Biologic therapy was used in 36.4% of CD patients and 9.1% of UC patients (P<0.001). CONCLUSION: In this population registry study, IBD prevalence was similar to that reported in other series in the region. A higher UC/CD ratio was observed due to the lower prevalence of CD compared to similar studies in South America.
Asunto(s)
Colitis Ulcerosa/epidemiología , Enfermedad de Crohn/epidemiología , Enfermedades Inflamatorias del Intestino/epidemiología , Adolescente , Adulto , Argentina/epidemiología , Femenino , Humanos , Masculino , Persona de Mediana Edad , Fenotipo , Sistema de Registros , Adulto JovenRESUMEN
When we read printed text, we are continuously predicting upcoming words to integrate information and guide future eye movements. Thus, the Predictability of a given word has become one of the most important variables when explaining human behaviour and information processing during reading. In parallel, the Natural Language Processing (NLP) field evolved by developing a wide variety of applications. Here, we show that using different word embeddings techniques (like Latent Semantic Analysis, Word2Vec, and FastText) and N-gram-based language models we were able to estimate how humans predict words (cloze-task Predictability) and how to better understand eye movements in long Spanish texts. Both types of models partially captured aspects of predictability. On the one hand, our N-gram model performed well when added as a replacement for the cloze-task Predictability of the fixated word. On the other hand, word embeddings were useful to mimic Predictability of the following word. Our study joins efforts from neurolinguistic and NLP fields to understand human information processing during reading to potentially improve NLP algorithms.
RESUMEN
ABSTRACT Objective: To define whether the electroneurophysiological stimulation would be a safe method for reducing injuries in nerve roots during surgery of lumbar spine arthrodesis, as well as verify whether there is a direct correlation between the intraoperative impedance values and the distance from the medial cortical pedicle screw. Methods: Randomized retrospective multicenter study of 10 patients who underwent arthrodesis of lumbar spine after conservative treatment failure, with a total of 50 pedicle screws instrumented. Reliable and safe impedance values were measured in order to reduce the risk of injury to nerve roots in the perioperative period, and these values were compared with the distance between the screw and the medial cortical of the pedicle by CT scan, measured in the immediate post-operative period. Results: There is no direct correlation between the intraoperative impedance values and the distance from the screw to the medial cortical of the pedicle. Conclusion: The electroneurostimulation proved to be a reliable quantitative method to reduce the risk of injury to nerve roots during surgery of lumbar spine arthrodesis when the measured values are greater than 10mA.
RESUMO Objetivo: Definir se a estimulação eletroneurofisiológica seria um método seguro para redução de lesões em raízes nervosas no intraoperatório de artrodese de coluna lombar, bem como verificar se há correlação direta entre os valores de impedância transoperatória e a distância do parafuso e a cortical medial do pedículo. Métodos: Estudo retrospectivo, multicêntrico com seleção randômica consecutiva de 10 pacientes que realizaram artrodese de coluna lombossacra após falha de tratamento conservador, com um total de 50 parafusos pediculares instrumentados. Os valores confiáveis e seguros de impedância para diminuir os riscos de lesão em raízes nervosas foram aferidos no período perioperatório, sendo esses valores comparados com os da distância entre o parafuso e a cortical medial do pedículo na tomografia computadorizada, aferida no pós-operatório imediato. Resultados: Não há correlação direta entre os valores de impedância transoperatória e a distância do parafuso até a cortical medial do pedículo. Conclusão: A eletroneuroestimulação mostrou ser um método quantitativo seguro para diminuir os riscos de lesões em raízes nervosas no intraoperatório de artrodese de coluna lombar quando os valores aferidos são maiores que 10mA.
RESUMEN Objetivo: Definir si la estimulación electroneurofisiológica sería un método seguro para reducir las lesiones en las raíces nerviosas durante la cirugía de artrodesis de columna lumbar, así como verificar si hay correlación directa entre los valores de impedancia perioperatoria y la distancia entre tornillo y la cortical medial del pedículo. Métodos: Estudio retrospectivo, multicéntrico, con selección aleatoria consecutiva de 10 pacientes que fueron sometidos a la artrodesis de columna lumbosacra después de falla del tratamiento conservador, con un total de 50 tornillos pediculares instrumentados. Valores fiables y seguros de impedancia para reducir el riesgo de lesiones a las raíces nerviosas se midieron en el periodo perioperatorio, y estos valores se compararon con la distancia entre el tornillo y la cortical medial del pedículo en la tomografía computarizada, medida en el período postoperatorio inmediato. Resultados: No existe una correlación directa entre los valores de impedancia perioperatoria y la distancia desde el tornillo hasta la cortical medial del pedículo. Conclusión: La electroneuroestimulación mostró ser un método cuantitativo fiable para reducir el riesgo de lesiones a raíces nerviosas durante la cirugía de artrodesis de columna lumbar cuando los valores medidos son mayores que 10mA.
Asunto(s)
Humanos , Vértebras Lumbares , Artrodesis , Impedancia Eléctrica , Estimulación Eléctrica , Raíces Nerviosas EspinalesRESUMEN
The MetaCyc database (MetaCyc.org) is a freely accessible comprehensive database describing metabolic pathways and enzymes from all domains of life. The majority of MetaCyc pathways are small-molecule metabolic pathways that have been experimentally determined. MetaCyc contains more than 2400 pathways derived from >46,000 publications, and is the largest curated collection of metabolic pathways. BioCyc (BioCyc.org) is a collection of 5700 organism-specific Pathway/Genome Databases (PGDBs), each containing the full genome and predicted metabolic network of one organism, including metabolites, enzymes, reactions, metabolic pathways, predicted operons, transport systems, and pathway-hole fillers. The BioCyc website offers a variety of tools for querying and analyzing PGDBs, including Omics Viewers and tools for comparative analysis. This article provides an update of new developments in MetaCyc and BioCyc during the last two years, including addition of Gibbs free energy values for compounds and reactions; redesign of the primary gene/protein page; addition of a tool for creating diagrams containing multiple linked pathways; several new search capabilities, including searching for genes based on sequence patterns, searching for databases based on an organism's phenotypes, and a cross-organism search; and a metabolite identifier translation service.
Asunto(s)
Bases de Datos de Compuestos Químicos , Enzimas/metabolismo , Redes y Vías Metabólicas , Bases de Datos Genéticas , Transporte de Electrón , Genoma , Internet , Redes y Vías Metabólicas/genética , Programas InformáticosRESUMEN
MOTIVATION: Key problems for computational genomics include discovering novel pathways in genome data, and discovering functional interaction partners for genes to define new members of partially elucidated pathways. RESULTS: We propose a novel method for the discovery of subsystems from annotated genomes. For each gene pair, a score measuring the likelihood that the two genes belong to a same subsystem is computed using genome context methods. Genes are then grouped based on these scores, and the resulting groups are filtered to keep only high-confidence groups. Since the method is based on genome context analysis, it relies solely on structural annotation of the genomes. The method can be used to discover new pathways, find missing genes from a known pathway, find new protein complexes or other kinds of functional groups and assign function to genes. We tested the accuracy of our method in Escherichia coli K-12. In one configuration of the system, we find that 31.6% of the candidate groups generated by our method match a known pathway or protein complex closely, and that we rediscover 31.2% of all known pathways and protein complexes of at least 4 genes. We believe that a significant proportion of the candidates that do not match any known group in E.coli K-12 corresponds to novel subsystems that may represent promising leads for future laboratory research. We discuss in-depth examples of these findings. AVAILABILITY: Predicted subsystems are available at http://brg.ai.sri.com/pwy-discovery/journal.html. CONTACT: lferrer@ai.sri.com SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
Asunto(s)
Escherichia coli K12/genética , Estudios de Asociación Genética/métodos , Redes y Vías Metabólicas/genética , Algoritmos , Biología Computacional , Genoma , Programas InformáticosRESUMEN
BACKGROUND: Genome context methods have been introduced in the last decade as automatic methods to predict functional relatedness between genes in a target genome using the patterns of existence and relative locations of the homologs of those genes in a set of reference genomes. Much work has been done in the application of these methods to different bioinformatics tasks, but few papers present a systematic study of the methods and their combination necessary for their optimal use. RESULTS: We present a thorough study of the four main families of genome context methods found in the literature: phylogenetic profile, gene fusion, gene cluster, and gene neighbor. We find that for most organisms the gene neighbor method outperforms the phylogenetic profile method by as much as 40% in sensitivity, being competitive with the gene cluster method at low sensitivities. Gene fusion is generally the worst performing of the four methods. A thorough exploration of the parameter space for each method is performed and results across different target organisms are presented. We propose the use of normalization procedures as those used on microarray data for the genome context scores. We show that substantial gains can be achieved from the use of a simple normalization technique. In particular, the sensitivity of the phylogenetic profile method is improved by around 25% after normalization, resulting, to our knowledge, on the best-performing phylogenetic profile system in the literature. Finally, we show results from combining the various genome context methods into a single score. When using a cross-validation procedure to train the combiners, with both original and normalized scores as input, a decision tree combiner results in gains of up to 20% with respect to the gene neighbor method. Overall, this represents a gain of around 15% over what can be considered the state of the art in this area: the four original genome context methods combined using a procedure like that used in the STRING database. Unfortunately, we find that these gains disappear when the combiner is trained only with organisms that are phylogenetically distant from the target organism. CONCLUSIONS: Our experiments indicate that gene neighbor is the best individual genome context method and that gains from the combination of individual methods are very sensitive to the training data used to obtain the combiner's parameters. If adequate training data is not available, using the gene neighbor score by itself instead of a combined score might be the best choice.
Asunto(s)
Genómica/métodos , Calibración , Bases de Datos Genéticas , Perfilación de la Expresión Génica/métodos , Genoma , Genómica/normas , Familia de Multigenes , FilogeniaRESUMEN
O consumo de álcool durante a gestação é importante problema de saúde pública, devido aos comprovados efeitos de toxicidade e teratogenicidade fetal a ele relacionado. Em decorrência da alta tolerância e disponibilidade para o consumo de álcool na sociedade, a freqüencia e a intensidade do uso pode tornar-se um fator de risco de grandes dimensões. O abuso do álcool está associado, de forma dose-dependente, a alterações na fertilidade, restrição do crescimento e desenvolvimento fetal, deficiências cognitivas, aumento da mortalidade, malformações congênitas, complicações no parto e problemas durante a infância. A síndrome alcoólica fetal (SAF) é uma condição irreversível caracterizada por anomalias crânio-faciais típicas, deficiência de crescimento, disfunções do sistema nervoso central e várias malformações associadas. O termo "efeitos fetais do álcool" foi proposto para um grupo de crianças expostas ao álcool intra-útero, mas que não possuíam o quadro clínico completo de SAF. O presente estudo tem o objetivo de realizar uma revisão bibliográfica dos principais aspectos do álcool relacionado à gestação, devido à grande prevalência e à gravidade dos problemas causados por essa associação
The alcohol consumption during pregnancy is an important public health problem, because of the proven toxic effects and fetal teratogenic related. In virtue the high tolerance and availability for the consumption of alcohol in the society, can make the frequency and the intensity of its use can become a factor of risk of great extents. The abuse of is associated, with a form dose-dependent, with changes in the fertiligy, restriction of the growth and fetal development, cognitive disabilities, increase of mortality, congenital malformations, labour complications and problems during the childhood. Fetal Alcohol Syndrome (FAS) is a nonreversible condition characterized by typical cranialface anomalies, disability of growth, abnormalities of the central nervous system and several malformations associate. The present study has the objective to make a bibliographic revision of the main aspects alcohol related with pregnancy, taking into consideration the great prevalence and gravity of the problems caused for this association