Digitala Vetenskapliga Arkivet

Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Evaluating a Commercial Deep Learning Model for Digitising Handwritten Healthcare Reports
Uppsala University, Disciplinary Domain of Science and Technology, Mathematics and Computer Science, Department of Information Technology.
2024 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Although many countries in the world have transitioned their reporting from analogue to digital, several still use handwritten healthcare reports. These often need to be manually converted in to a digital format, but with the recent rise of deep learning models, several academic models have been shown adept at automatically extracting information from documents. However, it is difficult to find any recent articles that evaluate a commercial solution. One such system is Microsoft Azure AI Document Intelligence, which allows the user to train a deep learning model for specific documents, in order to automatically extract the required information. In this thesis, four models were trained using Microsoft's system and tested on real healthcare reports from Peru. The ground truths to the reports were created in conjunction with Spanish-speaking pharmacologists. The results showed that when the variable names did not match the language of the report, the performance worsened significantly. Additionally, training with more than the required 5 training reports improved the results of the model. The best model achieved almost perfect results for checkboxes, was exactly correct on shorter text fields more often than not, and the median proportional Levenshtein distance was 11% for longer text fields.

Place, publisher, year, edition, pages
2024. , p. 42
Series
IT ; mIA 24 001
National Category
Engineering and Technology
Identifiers
URN: urn:nbn:se:uu:diva-538178OAI: oai:DiVA.org:uu-538178DiVA, id: diva2:1896734
Supervisors
Examiners
Available from: 2024-09-11 Created: 2024-09-11 Last updated: 2024-10-03Bibliographically approved

Open Access in DiVA

fulltext(7688 kB)266 downloads
File information
File name FULLTEXT01.pdfFile size 7688 kBChecksum SHA-512
66572a4b6586a0dcf3ac9de0d49b0ca2cb11382b1a31cc962a6a78bd4c9cbc23e91ca09069d40e5028c351d4be1f117c5fd688e0ed9ff560238b9f192a195a39
Type fulltextMimetype application/pdf

By organisation
Department of Information Technology
Engineering and Technology

Search outside of DiVA

GoogleGoogle Scholar
Total: 270 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 482 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf