SciPoll 396: SOT/EUROTOX Debate: Is There a Role for Artificial Intelligence (AI) and Machine Learning (ML) in Risk Decisions?
What is your level of experience with conducting or interpreting results from AI- and ML-based toxicity prediction models?
Results
(130 Answers)
Answer Explanations 19
Mostly for prediction.
In my provison as chair of the CONTAM Panel of EFSA (Parma, Italy), I frequently use AI and ML in risk assessment.
I have as a professor been involved with HACCP and risk assessment since I returned to academia in about 1990, when computers and AI were just starting. So I still review a lot of research in One Health and Food Safety, that are aligned to risk and hazard analysis. My feeling is , when reading publications, that some authors are relying on machine generated answers , rather than logical reasoning, to estimate risks. This results in over or under estimation of risks. For instance SARS-1 was overestimated and SARS-3 underestimated in the literature.
Moderate
use of (Q)SAR for predicting mutagenicity of compounds
My experience was in managing programs to develop and apply such approaches with the goal to inform regulatory and other decisions and not to actually conduct and evaluate such studies
Been doing qsar for 20 years
I used these techniques myself, have read extensively in this area and have reviewed the work of others.
graph theory, nodal analysis, #D reconstruction, etc.
I have used numerous AI and ML programs and now consider them a waste of time.
I am really willing to learn more in AI & ML.
Working with all available QSAR models (DEREK, Sarah, OECD models, previously Topkat etc)
Experience based on reliability of ADS
Non-existant at time of retirement 9 yrs ago.
I have written an article (in press) on cancer artificial intelligent, where detailed future priorities were suggested to be focused on integration and analyses of a large body of isolated available data to identify numerous knowledge gaps on electrobiology of immune neuroplasticity in health or disease processes before giving isolated and disconnected data to AI which could complicate accurate risk assessment formulations.
SAB and BoD member for computational chemistry/toxicology company.
Do some consulting in the field. Work on algorithms in the social sciences.
I’ve had more recent experience in the last 3 years
Only reading about.