Explainable regression guides laboratory tests of malaria candidates
Source check: Source checked. Not independent scientific confirmation.
Finding: Piecewise-regression models learned activity relationships from 386 Open Source Malaria compounds and prioritized fragments for a commercial-library search. Researchers tested 22 selected compounds and followed three in concentration-response and human-cell cytotoxicity assays. A02 and A10 showed qualitatively consistent parasite-growth inhibition across the initial screen and follow-up curves. The third candidate, A22, has an unresolved screen-versus-curve discrepancy. These are modest-potency laboratory starting points, without established treatment efficacy or secure strongest-hit and selectivity claims.
Technology: modSAR supervised piecewise linear regression; SHAP
Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.
- AI contribution
- The fitted model learns fingerprint breakpoints and region-specific activity coefficients, prioritizing fragments and predicting activity to narrow 934 commercial compounds to 97, then 22 after additional chemical filters.
- Test and result
- Twenty-two model-prioritized commercial compounds underwent direct parasite-growth screening. A02 and A10 showed consistent qualitative inhibition across screen and follow-up; the paper additionally reports A22 concentration-response and human-cell cytotoxicity measurements, with its conflicting screen result retained.
- Limitation / still unknown
- Clinical efficacy, mechanism and independent replication are unestablished. A 22 shows only 18.6% mean inhibition at 10 micromolar in the initial screen despite a later reported IC 50 of 8.29 micromolar; the discrepancy is unresolved, so exact potency and selectivity are not represented as secure estimates.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://eprints.lse.ac.uk/121101/1/Cardoso_Silva_optimisation_based_modelling_published.pdf · https://doi.org/10.1016/j.artmed.2023.102700
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Primary author-institution-deposited published version read through indexed PDF text for Methods, fragment-to-test selection, experimental results and front metadata; matched third-party journal-PDF figure inspected separately, with access distinction and numerical inconsistency retained. Current PubMed notice metadata checked..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.