LangExtract is a from builders at Google that makes it simple to show messy, unstructured textual content into clear, structured knowledge by leveraging LLMs. Customers can present just a few few-shot examples together with a customized schema and get outcomes based mostly on that. It really works each with proprietary in addition to native LLMs (through Ollama).
A big quantity of knowledge in healthcare is unstructured, making it an excellent space the place a instrument like this may be helpful. Scientific notes are lengthy and filled with abbreviations and inconsistencies. Necessary particulars reminiscent of drug names, dosages, and particularly hostile drug reactions (ADRs) get buried within the textual content. Subsequently, for this text, I needed to see if LangExtract might deal with hostile drug response (ADR) detection in medical notes. Extra importantly, is it efficient? Let’s discover out on this article. Be aware that whereas LangExtract is an open-source challenge from builders at Google, it’s not an formally supported Google product.
Only a fast word: I’m solely displaying how LanExtract works. I’m not a physician, and this isn’t medical recommendation.
▶️ Here’s a detailed Kaggle pocket book to observe alongside.
Why ADR Extraction Issues
An Antagonistic Drug Response (ADR) is a dangerous, unintended outcome attributable to taking a drugs. These can vary from delicate uncomfortable side effects like nausea or dizziness to extreme outcomes which will require medical consideration.
Detecting them shortly is vital for affected person security and pharmacovigilance. The problem is that in medical notes, ADRs are buried alongside previous situations, lab outcomes, and different context. Because of this, detecting them is hard. Utilizing LLMs to detect ADRs is an ongoing space of analysis. Some current works have proven that LLMs are good at elevating crimson flags however not dependable. So, ADR extraction is an efficient stress take a look at for LangExtract, because the objective right here is to see if this library can spot the hostile reactions amongst different entities in medical notes like medicines, dosages, severity, and many others.
How LangExtract Works
Earlier than we bounce into utilization, let’s break down LangExtract’s workflow. It’s a easy three-step course of:
- Outline your extraction job by writing a transparent immediate that specifies precisely what you need to extract.
- Present just a few high-quality examples to information the mannequin in direction of the format and degree of element you anticipate.
- Submit your enter textual content, select the mannequin, and let LangExtract course of it. Customers can then overview the outcomes, visualize them, or move them straight into their downstream pipeline.
The official GitHub repository of the instrument has detailed examples spanning a number of domains, from entity extraction in Shakespeare’s Romeo & Juliet to remedy identification in medical notes and structuring radiology stories. Do test them out.
Set up
First we have to set up the LangExtract library. It’s at all times a good suggestion to do that inside a digital atmosphere to maintain your challenge dependencies remoted.
pip set up langextractFiguring out Antagonistic Drug Reactions in Scientific Notes with LangExtract & Gemini
Now let’s get to our use case. For this walkthrough, I’ll use Google’s Gemini 2.5 Flash mannequin. You can additionally use Gemini Professional for extra advanced reasoning duties. You’ll must first set your API key:
export LANGEXTRACT_API_KEY="your-api-key-here"▶️ Here’s a detailed Kaggle pocket book to observe alongside.
Step 1: Outline the Extraction Process
Let’s create our immediate for extracting medicines, dosages, hostile reactions, and actions taken. We will additionally ask for severity the place talked about.
immediate = textwrap.dedent("""
Extract remedy, dosage, hostile response, and motion taken from the textual content.
For every hostile response, embody its severity as an attribute if talked about.
Use precise textual content spans from the unique textual content. Don't paraphrase.
Return entities within the order they seem.""")
Subsequent, let’s present an instance to information the mannequin in direction of the right format:
# 1) Outline the immediate
immediate = textwrap.dedent("""
Extract situation, remedy, dosage, hostile response, and motion taken from the textual content.
For every hostile response, embody its severity as an attribute if talked about.
Use precise textual content spans from the unique textual content. Don't paraphrase.
Return entities within the order they seem.""")
# 2) Instance
examples = [
lx.data.ExampleData(
text=(
"After taking ibuprofen 400 mg for a headache, "
"the patient developed mild stomach pain. "
"They stopped taking the medicine."
),
extractions=[
lx.data.Extraction(
extraction_class="condition",
extraction_text="headache"
),
lx.data.Extraction(
extraction_class="medication",
extraction_text="ibuprofen"
),
lx.data.Extraction(
extraction_class="dosage",
extraction_text="400 mg"
),
lx.data.Extraction(
extraction_class="adverse_reaction",
extraction_text="mild stomach pain",
attributes={"severity": "mild"}
),
lx.data.Extraction(
extraction_class="action_taken",
extraction_text="They stopped taking the medicine"
)
]
)
]Step 2: Present the Enter and Run the Extraction
For the enter, I’m utilizing an actual medical sentence from the ADE Corpus v2 dataset on Hugging Face.
input_text = (
"A 27-year-old man who had a historical past of bronchial bronchial asthma, "
"eosinophilic enteritis, and eosinophilic pneumonia introduced with "
"fever, pores and skin eruptions, cervical lymphadenopathy, hepatosplenomegaly, "
"atypical lymphocytosis, and eosinophilia two weeks after receiving "
"trimethoprim (TMP)-sulfamethoxazole (SMX) therapy."
)Subsequent, let’s run LangExtract with the Gemini-2.5-Flash mannequin.
outcome = lx.extract(
text_or_documents=input_text,
prompt_description=immediate,
examples=examples,
model_id="gemini-2.5-flash",
api_key=LANGEXTRACT_API_KEY
)Step 3: View the Outcomes
You may show the extracted entities with positions
print(f"Enter: {input_text}n")
print("Extracted entities:")
for entity in outcome.extractions:
position_info = ""
if entity.char_interval:
begin, finish = entity.char_interval.start_pos, entity.char_interval.end_pos
position_info = f" (pos: {begin}-{finish})"
print(f"• {entity.extraction_class.capitalize()}: {entity.extraction_text}{position_info}")
LangExtract accurately identifies the hostile drug response with out complicated it with the affected person’s pre-existing situations, which is a key problem in one of these job.
If you wish to visualize it, it’s going to create this .jsonl file. You may load that .jsonl file by calling the visualization perform, and it’ll create an HTML file for you.
lx.io.save_annotated_documents(
[result],
output_name="adr_extraction.jsonl",
output_dir="."
)
html_content = lx.visualize("adr_extraction.jsonl")
# Show the HTML content material straight
show((html_content))
Working with longer medical notes
Actual medical notes are sometimes for much longer than the instance proven above. As an example, right here is an precise word from the ADE-Corpus-V2 dataset launched below the MIT License. You may entry it on Hugging Face or Zenodo.
To course of longer texts with LangExtract, you retain the identical workflow however add three parameters:
extraction_passes runs a number of passes over the textual content to catch extra particulars and enhance recall.
max_workers controls parallel processing so bigger paperwork will be dealt with quicker.
max_char_buffer splits the textual content into smaller chunks, which helps the mannequin keep correct even when the enter could be very lengthy.
outcome = lx.extract(
text_or_documents=input_text,
prompt_description=immediate,
examples=examples,
model_id="gemini-2.5-flash",
extraction_passes=3,
max_workers=20,
max_char_buffer=1000
)Right here is the output. For brevity, I’m solely displaying a portion of the output right here.
If you’d like, you can too move a doc’s URL on to the text_or_documents parameter.
Utilizing LangExtract with Native fashions through Ollama
LangExtract isn’t restricted to proprietary APIs. You may as well run it with native fashions by way of Ollama. That is particularly helpful when working with privacy-sensitive medical knowledge that may’t depart your safe atmosphere. You may arrange Ollama domestically, pull your most well-liked mannequin, and level LangExtract to it. Full directions can be found within the official docs.
Conclusion
For those who’re constructing an data retrieval system or any software involving metadata extraction, LangExtract can prevent a major quantity of preprocessing effort. In my ADR experiments, LangExtract carried out nicely, accurately figuring out medicines, dosages, and reactions. What I observed is that the output straight relies on the standard of the few-shot examples supplied by the person, which implies whereas LLMs do the heavy lifting, people nonetheless stay an vital a part of the loop. The outcomes had been encouraging, however since medical knowledge is high-risk, broader and extra rigorous testing throughout numerous datasets remains to be wanted earlier than transferring towards manufacturing use.
