An electronic nose-based assistive diagnostic prototype for lung cancer detection with conformal prediction

Zhan, Xianghao; Zhan, Wang; Yue, Meng; Luo, Zhiyuan; Wang, You; Li, Guang

doi:10.1016/j.measurement.2020.107588

Cited by 42 publications

(28 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The part of unlabeled notes might still contain meaningful information related to classification. Such methods as semi-supervised learning and conformal predictions [30, 31] might be hold potential to make use of these unlabeled data, which could potentially further improve the prediction performance. Thirdly, this work focused on the prediction of ICD-10 codes and the structured codes was not tested in downstream tasks such as phenotyping or outcome prediction with machine learning.…”

Section: Discussionmentioning

confidence: 99%

Structuring clinical text with AI: old vs. new natural language processing techniques evaluated on eight common cardiovascular diseases

Zhan¹,

Humbert-Droz

Mukherjee

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Mining the structured data in electronic health records(EHRs) enables many clinical applications while the information in free-text clinical notes often remains untapped. Free-text notes are unstructured data harder to use in machine learning while structured diagnostic codes can be missing or even erroneous. To improve the quality of diagnostic codes, this work extracts structured diagnostic codes from the unstructured notes concerning cardiovascular diseases. Five old and new word embeddings were used to vectorize over 5 million progress notes from Stanford EHR and logistic regression was used to predict eight ICD-10 codes of common cardiovascular diseases. The models were interpreted by the important words in predictions and analyses of false positive cases. Trained on Stanford notes, the model transferability was tested in the prediction of corresponding ICD-9 codes of the MIMIC-III discharge summaries. The word embeddings and logistic regression showed good performance in the diagnostic code extraction with TF-IDF as the best word embedding model showing AU-ROC ranging from 0.9499 to 0.9915 and AUPRC ranging from 0.2956 to 0.8072. The models also showed transferability when tested on MIMIC-III data set with AUROC ranging from 0.7952 to 0.9790 and AUPRC ranging from 0.2353 to 0.8084. Model interpretability was showed by the important words with clinical meanings matching each disease. This study shows the feasibility to accurately extract structured diagnostic codes, impute missing codes and correct erroneous codes from free-text clinical notes with interpretable models for clinicians, which helps improve the data quality of diagnostic codes for information retrieval and downstream machine-learning applications.

show abstract

Section: Discussionmentioning

confidence: 99%

Structuring clinical text with AI: old vs. new natural language processing techniques evaluated on eight common cardiovascular diseases

Zhan¹,

Humbert-Droz

Mukherjee

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…One of the most important advantages of ANN is its ability to generalize information obtained from previously unknown data. In addition, neural networks are characterized by a high tolerance to additional perturbations, lack of continuity and deficiencies in the training set (125) Zahn (126), in his study, introduced a new application to predict the diagnosis of lung cancer with an e-nose. Samples were obtained from 31 patients aged 30 to 80 years with diagnosed lung cancer (adenocarcinoma, small cell carcinoma, squamous cell carcinoma, large cell carcinoma).…”

Section: Artificial Intelligence Contribution In An Artificialmentioning

confidence: 99%

What is the “Status questionis” of the e-nose. The future of artificial nose

2021

J B & Bio Engine;

View full text Add to dashboard Cite

An artificial nose (e-nose) is a multipotential electronic device, based on various sensors with the ability to recognize different odours, in the same way that the human olfactory sense does. An updated e-nose system will allow us to detect different oncological and/or degenerative diseases of the human being that today are diagnosed late. Other options would include providing specific information on the quality and condition of food, analysing and detecting the degree of environmental pollution, analysing perfumes and their essences, determining the composition and characteristics of certain beverages such as wine, tea, oil, cocoa and other products. The application of new technologies, such as artificial intelligence and nanotechnology, make it easier to distinguish many different odours in less time. In this paper, we have made a current investigation of the different types of e-nose existing today. Aim of the study Currently nanotechnology, artificial intelligence and computer science are tools that have revolutionary possibilities for the construction of a new e-nose device and progress over the current scopes. If we manage to unite the advantages provided by each of these new technologies that we have mentioned, we will be able to build a very useful device applicable in various fields such as health, food and beverage industry, perfumes and environment. Therefore, the objectives we have in this study are the following: To learn where artificial nose technology stands and what has been developed so far. Based on these findings, ask ourselves the following question: Is it possible to achieve an effective functioning or is it an unreachable project? To compile the majority of scientific articles published mainly in the last 10 years with examples of the use that has been made of artificial noses in the measurement of volatile compounds in the different fields mentioned. In the same way to gather the information published in the last years in relation to the usefulness, existence in the market and purposes of equipment that can measure the olfaction in the human being, what we will call the Smell-o-meter or olfactometer for human use. Material and Methods In the first part of this research we will gather most of the information existing so far in the international bibliography, as well as the achievements and utilities obtained to date. Following we will analyse all the new concepts related to e-nose devices that exist on sensors, gas chromatography, nanotechnology application, electronic engineering, materials and techniques as preliminary ideas for the development of the devices.

show abstract

“…Dado a los avances de este tipo de equipos electrónicos en cuanto a la mejora en cada uno de los sub-sistemas principales tales como, sistema de concentración, sistema de medida y procesado, hoy en día no solo se destaca su aplicabilidad al sector de la agroindustria y medio ambiente, sino al diagnóstico de diferentes tipos de enfermedades. Por ejemplo, para la detección de la enfermedad pulmonar obstructiva crónica o EPOC [8], para la detección del cáncer de pulmón [9], entre otros [10].…”

Section: Aplicaciones De Los Sistemas De Olfato Electrónicounclassified

Análisis de volátiles en el proceso de fermentado de cacao, mediante una nariz electrónica para el control de calidad del producto en Norte de Santander-Cúcuta

2020

View full text Add to dashboard Cite

El presente estudio consiste de una nariz electrónica compuesta de 10 sensores de gases de tipo MQ para la clasificación de muestras de CLON ICS-95 de cacao. El desarrollo de las pruebas fue de tipo cualitativo, obteniendo una huella digital que caracterizó cada clase, las cuales fueron: Fermentado deseado: 144 horas, sobre-fermentado y mala fermentación cacao infectado con monilia. Todos los sensores usados en las diferentes pruebas fueron de material de óxidos metálicos con capacidad de medir diversos tipos de gases, butanos, propanos, alcoholes, monóxido de carbono en diferentes concentraciones, donde al hacer contacto con los volátiles asociados producen una alteración en el voltaje de salida. Las señales se adquirieron mediante un sistema de adquisición de datos basado en tarjeta Arduino y uso del software Labview, permitiendo el almacenamiento de los datos. El algoritmo para la extracción de parámetros, pre-procesamiento y procesamiento de datos se realizó mediante el uso de software Python. Los resultados se analizaron implementando análisis de componente principales PCA y ejecución de dos métodos de pre-procesamiento de datos, como el centrado y escalado de datos, logrando un porcentaje de varianza en los componentes principales de 97.8% y con el método Manhattan se obtuvo un 93.8% del porcentaje de varianza en la componente principal PC1. Con estos resultados se logró observar que el sistema de olfato electrónico fue capaz de clasificar los datos de acuerdo a las clases definidas, fermentado deseado: 144 horas, sobre-fermentado y mala fermentación cacao infectado con monilia

show abstract

An electronic nose-based assistive diagnostic prototype for lung cancer detection with conformal prediction

Cited by 42 publications

References 43 publications

Structuring clinical text with AI: old vs. new natural language processing techniques evaluated on eight common cardiovascular diseases

Structuring clinical text with AI: old vs. new natural language processing techniques evaluated on eight common cardiovascular diseases

What is the “Status questionis” of the e-nose. The future of artificial nose

Análisis de volátiles en el proceso de fermentado de cacao, mediante una nariz electrónica para el control de calidad del producto en Norte de Santander-Cúcuta

Contact Info

Product

Resources

About