The Center for Suicide Research and Prevention is seeking Fall 2025 student interns!
💻 Hybrid (in-person & virtual)
📅 Start: Fall 2025
🌍 Open to all students – no prior research experience required
🔗 Learn more: https://t.co/j6dKpGpNr3
📝 Apply now: https://t.co/UtlscltCpR
Our new paper on continuous-time, dynamic suicide attempt risk prediction is now published online at @npjDigitalMed.
Using a large-scale electronic health records database, we adapt neural ODEs—naturally suited for modeling evolving latent states like suicidal tendencies and handling irregularly-spaced temporal data such as healthcare records—to address two key limitations of existing methods: static risk scoring and reliance on healthcare visits, which create critical gaps in risk assessment (figure).
Thank you and @jorsmo, @bwang482, and @MGHPrecisionPsy for the amazing support and teamwork!
🚨Come work with us @UniofOxford on new data science techniques 👩💻🧑💻 to help predict and treat episodes of severe mental illness 🧠
Deadline: March 3.
Postdoc in data science: https://t.co/axphq5TIKG
Research assistants (2 posts): https://t.co/V0nzMtodM9
We have an exciting lineup of speakers for this year's 4th Annual Conference on Precision Psychiatry!
Students, fellows, interns, and trainees can register for free!
Register today at https://t.co/SUlS4AGz9w
Questions can be directed to [email protected]
@ruima Contrarian not; acquisition of specific social skills is very much valued by schools with sufficient resources, e.g. the British public schools (in both good and bad way). The real question is how socialization (of children) can be guided and improved with little resources.
@rich_lynna31339@jschunter@alexjc I have read the reporting in Chinese and the ruling does say "防范程度应达到:用户正常使用与奥特曼相关的提示词,不能生成与案涉奥特曼作品实质性相似的图片" which essentially means measures have to be taken so Ultraman-related content cannot be generated by its users
@jschunter@alexjc Is the correct interpretable from this ruling that if you make your model available to your users, you need to make sure infringing output cannot be generated by any prompt one can think of?