Machine Learning Applied to COVID-19: A Review of the Initial Pandemic Period
Abstract Diagnostic and decision-making processes in the 2019 Coronavirus treatment have combined new standards using patient chest images, clinical and laboratory data. This work presents a systematic review aimed at studying the Artificial Intelligence (AI) approaches to the patients’ diagnosis or...
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2023-05-01
|
| Series: | International Journal of Computational Intelligence Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44196-023-00236-3 |
| _version_ | 1827952526493745152 |
|---|---|
| author | Leandro Y. Mano Alesson M. Torres Andres Giraldo Morales Carla Cristina P. Cruz Fabio H. Cardoso Sarah Hannah Alves Cristiane O. Faria Regina Lanzillotti Renato Cerceau Rosa Maria E. M. da Costa Karla Figueiredo Vera Maria B. Werneck |
| author_facet | Leandro Y. Mano Alesson M. Torres Andres Giraldo Morales Carla Cristina P. Cruz Fabio H. Cardoso Sarah Hannah Alves Cristiane O. Faria Regina Lanzillotti Renato Cerceau Rosa Maria E. M. da Costa Karla Figueiredo Vera Maria B. Werneck |
| author_sort | Leandro Y. Mano |
| collection | DOAJ |
| description | Abstract Diagnostic and decision-making processes in the 2019 Coronavirus treatment have combined new standards using patient chest images, clinical and laboratory data. This work presents a systematic review aimed at studying the Artificial Intelligence (AI) approaches to the patients’ diagnosis or evolution with Coronavirus 2019. Five electronic databases were searched, from December 2019 to October 2020, considering the beginning of the pandemic when there was no vaccine influencing the exploration of Artificial Intelligence-based techniques. The first search collected 839 papers. Next, the abstracts were reviewed, and 138 remained after the inclusion/exclusion criteria was performed. After thorough reading and review by a second group of reviewers, 64 met the study objectives. These papers were carefully analyzed to identify the AI techniques used to interpret the images, clinical and laboratory data, considering a distribution regarding two variables: (i) diagnosis or outcome and (ii) the type of data: clinical, laboratory, or imaging (chest computed tomography, chest X-ray, or ultrasound). The data type most used was chest CT scans, followed by chest X-ray. The chest CT scan was the only data type that was used for diagnosis, outcome, or both. A few works combine Clinical and Laboratory data, and the most used laboratory tests were C-reactive protein. AI techniques have been increasingly explored in medical image annotation to overcome the need for specialized manual work. In this context, 25 machine learning (ML) techniques with a highest frequency of usage were identified, ranging from the most classic ones, such as Logistic Regression, to the most current ones, such as those that explore Deep Learning. Most imaging works explored convolutional neural networks (CNN), such as VGG and Resnet. Then transfer learning which stands out among the techniques related to deep learning has the second highest frequency of use. In general, classification tasks adopted two or three datasets. COVID-19 related data is present in all papers, while pneumonia is the most common non-COVID-19 class among them. |
| first_indexed | 2024-04-09T14:00:05Z |
| format | Article |
| id | doaj.art-80b372557a2e4e9c9fdd64b9236b8f96 |
| institution | Directory Open Access Journal |
| issn | 1875-6883 |
| language | English |
| last_indexed | 2024-04-09T14:00:05Z |
| publishDate | 2023-05-01 |
| publisher | Springer |
| record_format | Article |
| series | International Journal of Computational Intelligence Systems |
| spelling | doaj.art-80b372557a2e4e9c9fdd64b9236b8f962023-05-07T11:23:24ZengSpringerInternational Journal of Computational Intelligence Systems1875-68832023-05-0116112410.1007/s44196-023-00236-3Machine Learning Applied to COVID-19: A Review of the Initial Pandemic PeriodLeandro Y. Mano0Alesson M. Torres1Andres Giraldo Morales2Carla Cristina P. Cruz3Fabio H. Cardoso4Sarah Hannah Alves5Cristiane O. Faria6Regina Lanzillotti7Renato Cerceau8Rosa Maria E. M. da Costa9Karla Figueiredo10Vera Maria B. Werneck11Department of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityDepartment of Informatics and Computer Science, Rio de Janeiro State UniversityAbstract Diagnostic and decision-making processes in the 2019 Coronavirus treatment have combined new standards using patient chest images, clinical and laboratory data. This work presents a systematic review aimed at studying the Artificial Intelligence (AI) approaches to the patients’ diagnosis or evolution with Coronavirus 2019. Five electronic databases were searched, from December 2019 to October 2020, considering the beginning of the pandemic when there was no vaccine influencing the exploration of Artificial Intelligence-based techniques. The first search collected 839 papers. Next, the abstracts were reviewed, and 138 remained after the inclusion/exclusion criteria was performed. After thorough reading and review by a second group of reviewers, 64 met the study objectives. These papers were carefully analyzed to identify the AI techniques used to interpret the images, clinical and laboratory data, considering a distribution regarding two variables: (i) diagnosis or outcome and (ii) the type of data: clinical, laboratory, or imaging (chest computed tomography, chest X-ray, or ultrasound). The data type most used was chest CT scans, followed by chest X-ray. The chest CT scan was the only data type that was used for diagnosis, outcome, or both. A few works combine Clinical and Laboratory data, and the most used laboratory tests were C-reactive protein. AI techniques have been increasingly explored in medical image annotation to overcome the need for specialized manual work. In this context, 25 machine learning (ML) techniques with a highest frequency of usage were identified, ranging from the most classic ones, such as Logistic Regression, to the most current ones, such as those that explore Deep Learning. Most imaging works explored convolutional neural networks (CNN), such as VGG and Resnet. Then transfer learning which stands out among the techniques related to deep learning has the second highest frequency of use. In general, classification tasks adopted two or three datasets. COVID-19 related data is present in all papers, while pneumonia is the most common non-COVID-19 class among them.https://doi.org/10.1007/s44196-023-00236-3COVID-19Machine learningImage dataClinical and laboratorial dataDiagnosis/outcome |
| spellingShingle | Leandro Y. Mano Alesson M. Torres Andres Giraldo Morales Carla Cristina P. Cruz Fabio H. Cardoso Sarah Hannah Alves Cristiane O. Faria Regina Lanzillotti Renato Cerceau Rosa Maria E. M. da Costa Karla Figueiredo Vera Maria B. Werneck Machine Learning Applied to COVID-19: A Review of the Initial Pandemic Period International Journal of Computational Intelligence Systems COVID-19 Machine learning Image data Clinical and laboratorial data Diagnosis/outcome |
| title | Machine Learning Applied to COVID-19: A Review of the Initial Pandemic Period |
| title_full | Machine Learning Applied to COVID-19: A Review of the Initial Pandemic Period |
| title_fullStr | Machine Learning Applied to COVID-19: A Review of the Initial Pandemic Period |
| title_full_unstemmed | Machine Learning Applied to COVID-19: A Review of the Initial Pandemic Period |
| title_short | Machine Learning Applied to COVID-19: A Review of the Initial Pandemic Period |
| title_sort | machine learning applied to covid 19 a review of the initial pandemic period |
| topic | COVID-19 Machine learning Image data Clinical and laboratorial data Diagnosis/outcome |
| url | https://doi.org/10.1007/s44196-023-00236-3 |
| work_keys_str_mv | AT leandroymano machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT alessonmtorres machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT andresgiraldomorales machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT carlacristinapcruz machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT fabiohcardoso machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT sarahhannahalves machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT cristianeofaria machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT reginalanzillotti machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT renatocerceau machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT rosamariaemdacosta machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT karlafigueiredo machinelearningappliedtocovid19areviewoftheinitialpandemicperiod AT veramariabwerneck machinelearningappliedtocovid19areviewoftheinitialpandemicperiod |