Hospital documents, which can sometimes be hundreds of pages long, are integrated to your Anthuria account from your EHR. Anthuria AI extracts ICD-10 codes automatically based on relevant evidence within and across documents. Documents are made searchable and easy to navigate with Anthuria's Condition Search. This can save you significant time that would otherwise be spent manually scanning documents for key information.
ICD-10 Code AI Agent
ICD-10 Extraction
Patient View uses AI to automatically scan and extract ICD-10 codes from a patient's hospital records.
The application will show you a list of all ICD-10 codes that the AI has found.
For each code, Patient View identifies the document and specific text that served as the evidence for that code. This allows you to quickly verify the recommendation.
Toggle through the results easily with the toggle options on the left side bar or above the document being viewed.
Single Click Update to EHR
Update diagnoses in your EHR instantly from a patient’s ICD-10 code page—no extra steps required.
Use the "Add diagnosis in EHR" option from the ICD-10 code of a given patient to go directly to the Diagnosis for that patient and update in realtime.
Download ICD-10 Codes
You can view the ICD-10 codes in a list or on a sidebar and even export the entire list to Excel for easy review.
You can agree with a code by giving it a thumbs-up 👍 or disagree with a thumbs-down 👎 and provide feedback. This feedback helps improve the AI model's accuracy. Your approval will also appear as a column on the export, allowing you and others to easily filter and sort accordingly.
Bulk Export: From the primary Patient View page you can bulk export ICD-10 codes across patients and facilities. You will be able to multi-select facilities to include, admittance date timeframe, and document time frame to include. The time frame is "Last 14 days" or "Across all time", using the last 14 days is how we recommend filtering out old hospital documents for readmitted residence (Last 14 days is the default).
ICD-10 Specificity
The specificity of the ICD-10 codes provided by Patient View is directly related to the detail found in the source documents. If a document only provides general information, the code will be more general (e.g., "unspecified"). However, if the document contains specific details, such as the exact location of a fracture, the AI will provide a more precise code.
Insight View vs. Original View
Patient View presents hospital documents in two ways to give you maximum flexibility.
Insight View: This view intelligently translates PDFs into a searchable "markdown" format. While the formatting may look slightly different from the original PDF, all the content is exactly the same and is now fully searchable. This is the primary view for most of your work in Patient View.
Original View: If you need to see the document in its original PDF format, you can easily toggle to the Original View. This is useful if you want to verify formatting or layout.
Condition Search
The Condition Search tab is where you can search for specific clinical conditions within a patient's hospital documents.
Preloaded Conditions: Patient View comes with a set of preloaded conditions that are often looked for, such as:
Isolation
COPD
Respiratory failure
Diabetes
Mechanically altered diet
IV fluids
Nutrition
Fall
Customization: You can easily customize these searches. If you click on a condition like "Diabetes," you'll see a list of search terms (e.g., "diabetes," "DM"). You can add or remove terms as needed.
Free Search: Need to find something specific that's not in the preloaded conditions? Use the free search bar to look for any word or phrase. You can also save a custom search as a permanent category for future use. For example, if you frequently need to check for a "history of homelessness," you can create a search group for it that looks for terms like "homeless" or "unhoused."
Toggle Through Results: Just like with ICD-10 codes, the Condition Search will highlight where a search term is found in the documents. You can quickly toggle through the results to see the information in context without having to read hundreds of pages.