December 5, 2023: MediDocHelper, an automated medical data processing tool from GenomiCare Biotechnology, has been re-released. The system uses generative AI technology to automate the analysis of medical data, transforming various forms of clinical information into structured reports, providing convenience for physicians while dramatically simplifying workflow and saving time and energy.
In today’s era of information explosion, automated information processing has become an inevitable trend. Especially in the medical field, a large amount of clinical information exists in unstructured forms such as images and text, and the lack of consistent formats and standards makes it difficult to extract and integrate the information for effective data analysis and mining, which in turn restricts the rapid progress of medical research and clinical decision-making. Therefore, overcoming the challenges of automated clinical information processing will bring more opportunities and innovations to the medical field.
GenomiCare Biotechnology responds to this challenge with the timely launch of MediDocHelper, an automated medical data processing tool equipped with powerful image-text recognition technology that can automatically convert text from images or scanned documents into editable text. In addition, MediDocHelper uses several world-class big language models, including OpenAI, Baidu Yiyan and ChatGLM, which have been trained and optimized many times to further improve their accuracy. In addition to optimizing the big language models, we have also improved the cue word templates to better automate the identification and extraction of key information from medical data into a structured data format, with the goal of helping healthcare organizations collect large amounts of clinical data more accurately, efficiently and customizably for subsequent analysis and use.
Platform Features:
- Image Text Recognition: It can quickly recognize and scan text information in images, and convert text in images or scans to editable text.
- Text content correction: automatic error correction and correction based on large language models.
- Text content recognition and classification: Combined with GenomiCare ‘s independently created clinical data organization cue word templates, the text content is intelligently classified according to the logical structure of paragraphs and chapters.
- Content extraction and structuring: Through powerful intelligent algorithms, key information is extracted and categorized from the large amount of text data, and the text content is transformed into structured data.
Figure 1: Image Text Recognition Results
- Editable text after image recognition, retaining the original format of the image, and easy to proofread against the top and bottom.
- The recognized text has been classified in terms of accuracy, where green background represents high accuracy, yellow represents medium accuracy, and red represents low accuracy
Figure 2: AI Proofreading
- Accurate Text Proofreading and Replacement automatically corrects errors and corrections through intelligent proofreading and provides the ability to quickly and directly replace them.
Figure 3: Extracting and Structuring Textual Content
- Automatically generate structured information by referring to the original text of the segmented paragraphs and configuring classification templates and large language models.
Automation tools can, to a certain extent, provide new solutions to existing problems in the medical industry, optimize and reconstruct some of the cumbersome and complicated work contents in the medical industry, and drive the medical industry to speed up efficiently. GenomiCare has a long way to go, and we believe that in the near future, with the continuous progress of information processing technology, we can open up more application scenarios and expand more ways and opportunities of medical information processing.