Advanced searches left 3/3

Artificial Intelligence - PLOS

Summarized by Plex Scholar
Last Updated: 10 January 2023

* If you want to update the article please login/register

Improving quality control in the routine practice for histopathological interpretation of gastrointestinal endoscopic biopsies using artificial intelligence

We therefore established a digital pathology total solution incorporating artificial intelligence classification coder models and pathology laboratory information for GI endoscopic biopsy samples in order to establish a fast quality control device that was used in clinical practice for a three-month trial conducted by four pathologists. Methods and results: Our entire slide image classification system was based on patch-generator, patch-level classifier, and a WSI-level classifier. The WSI classifier categorized histopathological WSIs of colorectal and gastric endoscopic biopsy specimens, respectively, into three classes in laboratory experiments. Pathologists reviewed nearly 710 slides compared to those in the traditional monthly QC trial period; over 80% of GI endoscopy biopsy slides were double-checked by the AI models during the 3-month AI-assisted daily QC trial run period.

Source link: https://doi.org/10.1371/journal.pone.0278542


Is artificial intelligence capable of generating hospital discharge summaries from inpatient records?

Medical professionals are heavily clerical, and artificial intelligence can effectively support physicians by generating clinical summaries. However, it's unclear if hospital discharge summaries can be generated automatically from patient records stored in electronic health information. Consequently, this research investigated the sources of evidence in discharge summaries. This report developed and annotated clinical role labels that represent the subjectivity of the expressions' subjectivity and inferential reasoning, which then automatically assigns them automatically. The following was revealed: The analysis revealed the following: The following information in the discharge summary came from other sources other than inpatient ones. Patient referral documents constituted 18% of the expressions derived from third-party reports, accounting for 43% of the patient's past medical records, according to the second, patientu2019s past clinical records, and patient referral documents constituted 18% of the expressions derived from external sources. Author summary: Medical research necessitates significant paperwork; therefore, automated processing of clinical data can reduce the burden on medical professionals. This review examined whether discharge summaries can be created automatically from inpatient data to facilitate further processing. Every piece of data in the discharge summaries is manually coded to determine if it came from inpatient records for this purpose. Patientu2019s past medical records and patient referral reports were included in these external sources. The authoring of discharge summaries must be considered infeasible by a wholly automated generation of discharge summaries, and future research efforts must be directed toward a semi-automated generation with the aim of increasing human-machine coordination in the production of discharge summaries.

Source link: https://doi.org/10.1371/journal.pdig.0000158


Assessing electrocardiogram changes after ischemic stroke with artificial intelligence

Methods: We obtained ECGs from a healthy population and patients with IS to be able to identify abnormalities in post-IS ECGs, then analysed participant demographics and ECG parameters to identify abnormalities in post-IS ECGs. Next, we trained the convolutional neural network, random forest, and support vector machine models to automatically detect the changes in the ECGs; in addition, we compared CNN scores of good prognosis and poor prognosis to determine the CNN's prognostic value of CNN's CNN model to determine the prognosis value of CNN's CNN scores to determine the prognosis value of CNN's CNN scores of good prognosis and poor prognosis We also used a gradient class activation map to localize the key abnormalities. After the first ischemic stroke and normal ECGs, the first ischemic stroke prompted the first ischemic stroke, and normal ECGs after the first ischemic stroke drew. The A-ISs and N-ISs obtained from all three models' forecast scores are statistically different from the N-Ns'. Grad-CAM reported that the V4 lead might have the highest likelihood of abnormality.

Source link: https://doi.org/10.1371/journal.pone.0279706

* Please keep in mind that all text is summarized by machine, we do not bear any responsibility, and you should always check original source before taking any actions

* Please keep in mind that all text is summarized by machine, we do not bear any responsibility, and you should always check original source before taking any actions