Connectionists: CFP: ToxHabits Shared Task (BioCreative IX at IJCAI2025): Named entity recognition of toxic habits (drug abuse, smoking & alcohol in clinical texts)

Thu May 15 11:15:09 EDT 2025

(Apologies for cross-posting)

CFP: ToxHabits Shared Task (BioCreative IX will be colocated with IJCAI 2025
<https://2025.ijcai.org/> in Montreal, CA)

Named entity recognition of toxic habits including substance abuse, smoking
& alcohol consumption in clinical texts

https://temu.bsc.es/toxhabits/

The ToxHabits track focuses on the automatic detection of toxic habit
entity mentions in clinical case reports written in Spanish. We also plan
to release a multilingual version of the dataset, including versions in
English and other languages.

Key information:

   -

   Web: https://temu.bsc.es/toxhabits/
   -

   Data: <https://doi.org/10.5281/zenodo.6408476>
   https://zenodo.org/records/15304758
   -

   BioCreative web: https://www.ncbi.nlm.nih.gov/research/bionlp/biocreative
   -

   Registration form (Track 3- ToxHabits):
   <https://temu.bsc.es/distemist/registration/>
   https://docs.google.com/forms/d/e/1FAIpQLSeHe9rLVGHGNtK0ExUFlNyGBJJ_VKPScx-N5ODBJbk3v5DoMQ/viewform

Motivation
Systems capable of detecting and characterizing toxic habits—including not
only smoking and alcohol consumption but also various forms of drug or
substance abuse—from medical texts are crucial for nearly all healthcare
data mining, medical AI, and predictive clinical applications. There is a
pressing need to automatically extract toxic habit information, especially
considering that, according to recent drug abuse statistics, in some
countries nearly half of individuals aged 12 and older have used illicit
drugs at least once. In the United States alone, drug overdose deaths since
2000 are approaching one million. These figures highlight a clear global
burden and the urgent need to improve the systematic extraction and
characterization of substance abuse information.

To address these challenges, we have proposed the ToxHabits track as part
of the upcoming BioCreative IX initiative, which will be held in
conjunction with the prestigious IJCAI 2025 international conference.

Automatic detection of toxic habits are key for a range of clinical use
cases and applications:

1. Risk Assessment for Chronic Diseases: Toxic habits are major risk
factors for cardiovascular disease, chronic obstructive pulmonary disease
(COPD), liver cirrhosis, and cancer. Knowing these habits aids in risk
stratification and preventive care planning.

2. Anesthesia and Surgical Planning: Substance use can affect anesthetic
response, bleeding risk, wound healing, and postoperative complications.
For instance, alcohol and smoking increase the risk of surgical site
infections.

3. Mental Health Diagnosis and Management: Substance abuse is closely
linked with depression, anxiety, psychosis, and suicidal behavior. Accurate
assessment is critical for correct psychiatric diagnosis and effective
treatment.

4. Medication Metabolism and Drug Interactions: Smoking (e.g., induces
CYP1A2) and alcohol can alter drug metabolism. Drug abuse may interact with
prescribed medications, leading to reduced efficacy or increased toxicity.

5. Prenatal and Maternal Care: In pregnant patients, toxic habits pose
risks for fetal alcohol syndrome, low birth weight, preterm birth, and
developmental disorders. Early detection is key to harm reduction.

6. Infectious Disease Screening and Prevention: Drug abuse, especially via
injection, increases the risk of HIV, hepatitis B/C, and other
transmissible infections. Alcohol abuse may contribute to high-risk
behaviors increasing disease transmission.

7. Liver Function and Hepatology: Chronic alcohol use is a major cause of
fatty liver, hepatitis, and cirrhosis. Identifying alcohol use is crucial
when evaluating liver enzyme abnormalities or planning hepatotoxic
treatments.

8. Oncology: Risk and Treatment Planning:Toxic habits are associated with
cancers of the lung, liver, pancreas, oral cavity, and others. Continuing
such habits during cancer treatment can affect prognosis and response.

9. Pain Management and Opioid Prescribing: In patients with a history of
substance abuse, opioid prescribing requires careful risk-benefit analysis
and often alternative pain control strategies.

10. Emergency and Acute Care Triage: In cases of trauma, overdose, or
altered mental status, knowledge of toxic habits can guide differential
diagnosis, antidote use, and monitoring strategies.

Inspired by previous initiatives (e.g., n2c2, CLEF, or TREC) and shared
tasks (e.g., Symptemist, PharmaCoNER, or DisTEMIST), we are launching the
ToxHabits shared task as part of the BioCreative 2025 evaluation
initiative, featuring the following three sub-tracks:

ToxNER (Subtask 1): Substance Mention Detection
Participants are required to return the starting and ending character
positions (spans) of each substance mention, along with its substance type
(i.e., Tobacco, Alcohol, Drug, or Cannabis).

ToxUse (Subtask 2): Substance Use/Consumption Characteristics Detection
Participants must return the starting and ending character positions
(spans) of substance use characteristics, along with the corresponding
label (i.e., Amount, Type, Frequency, Duration, Method, History).

Tentative schedule

   -

   ToxHabits registration: May 26th
   -

   ToxHabits Train set release: April 30th
   -

   Annotation Guidelines release: TBA
   -

   ToxHabits Test set release: May 26th
   -

   ToxHabits Test Prediction submission: June 1st (Anywhere on Earth)
   -

   ToxHabits results release: June 5th
   -

   IJCAI’25 Workshop Team Invitation: June 6th
   -

   BioCreative IX proceedings paper submission: June 16th
   -

   BioCreative IX proceedings paper reviews: July 1st
   -

   IJCAI 2025 August 16-22th (Montreal, Canada)

BioCreative proceedings and IJCAI workshop

Teams participating in ToxHabits will be invited to contribute a system
description paper to the BioCreative 2025 Working Notes proceedings and to
give a flash presentation of their approach during the BioCreative 2025
session. BioCreative IX will be co-located with IJCAI 2025 in Montreal,
Canada. The meeting will take place from August 16 to 22, 2025.
For more information, see: https://2025.ijcai.org/

Workshop Proceedings and Special Issue:

The BioCreative IX Proceedings will include short systems description
papers from participating teams and will be available by published online
by the time of the workshop. In addition, we are pleased to announce that,
following the tradition of previous BioCreative initiatives, top-performing
teams will be invited to submit a longer technical journal paper for
publication in a special issue. Invitations to submit will be sent after
the workshop.

All BioCreative IX tracks

Track 1: MedHopQA

Track 2: Sentence Segmentation of Real-Life Clinical Notes

*Track 3: ToxHabits

See: https://www.ncbi.nlm.nih.gov/research/bionlp/biocreative9

Organizers

   -

   Salvador Lima-López, Barcelona Supercomputing Center (BSC), Spain.
   -

   Wesam Al-Nabki, Barcelona Supercomputing Center (BSC), Spain.
   -

   Gabriel Vayá-Abad, Barcelona Supercomputing Center (BSC), Spain.
   -

   Eulàlia Farré-Maduell, Barcelona Supercomputing Center (BSC), Spain.
   -

   Jan Rodríguez Miret, Barcelona Supercomputing Center (BSC), Spain.
   -

   Martin Krallinger, Barcelona Supercomputing Center (BSC), Spain.

Scientific Committee

   -

   Dina Demner-Fushman, US National Institute of Health / National Library
   of Medicine
   -

   Ozlem Uzuner, George Mason University
   -

   Asma Ben Abacha, Microsoft Health AI
   -

   Anastasios Nentidis, National Center for Scientific Research Demokritos,
   Greece
   -

   Anastasia Krithara, National Center for Scientific Research Demokritos,
   Greece
   -

   Lourdes Araujo, Universidad Nacional de Educación a Distancia (UNED)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20250515/e13d02ba/attachment.html>