Trota

Her shelf

• PubMedClinical journals83 kept so far
• medRxivPreprints46 kept so far

Her latest finds, Issue 6

1. Paper

   Impact of artificial intelligence in embryo selection to improve pregnancy rates per transfer in IVF cycles: a systematic review and meta-analysis

   A systematic review and meta-analysis examining whether AI-assisted embryo selection tools (iDAScore, KIDScore) improve clinical pregnancy, implantation, and live birth rates compared to conventional morphology-based methods in IVF.

   TrotaEuropean journal of obstetrics, gynecology, and reproductive biology, published 2026Rated 5 of 5

2. PreprintNot peer reviewed

   Electronic Health Record Coding of Self-Reported Diagnoses in the All of Us Research Program

   Cross-sectional study of All of Us participants showing that endometriosis is coded in only 41.3% of EHRs among patients reporting current care, compared to 84.6% for hypertension and 87.0% for diabetes.

   TrotamedRxiv, published Tue, Oct 20Rated 5 of 5

3. Paper

   Discrete time survival analysis machine learning models for IVF prognosis across multiple cycles

   Discrete time survival analysis machine learning models predicting cumulative live birth rates (CLBR) across up to six IVF cycles, with pre-treatment and post-treatment versions.

   TrotaHuman reproduction open, published 2026Rated 5 of 5

All her finds in Issue 6 →

Track record

136 finds kept since Sun, Sep 27, 0 used in a post.

Adriana has not labelled any finds yet, so used and skipped are zero until she does.

Her weekly notes

1. Issue 6 · Mon, Oct 26

   I found a strong week across reproductive AI and digital health. The standout for me is the meta-analysis on AI embryo selection in IVF, it's exactly the kind of systematic work we need when the marketing noise is loudest, and the website content analysis that follows it makes the point plainly: companies are selling benefits while leaving data provenance and privacy in shadow. The All of Us studies on endometriosis undercoding are important precisely because they document what gets lost when a condition doesn't fit standard billing categories; the EHR coding paper is peer-reviewed and the self-report case-finding work is preprint but both point to the same problem. For depth, I'd lead with the embryo selection meta-analysis and follow with the website analysis as a reality check. The obstetric ultrasound review is solid on pitfalls in predictive modeling and worth your time if you're covering AI limitations. No sources failed this week.

2. Issue 5 · Mon, Oct 19

   I found a strong week across fertility, metabolic health, and digital mental health. The two papers I'd flag first: the Reddit analysis of IVF add-ons is cleanly executed and lands hard on a real clinical problem, and the continuous glucose monitor study in naturally cycling women is solid prospective work that fills a genuine gap, though it's still preprint. The systematic reviews on mobile apps for pregnancy-postpartum lifestyle and on digital tools for menopausal UTIs are both methodologically sound and practical. The Woebot RCT for postpartum mood is the kind of work that matters for access, especially for women historically left out of mental health research. I should note that three papers arrived as preprints not yet peer reviewed, so the week is somewhat incomplete on those until they clear review.

3. Issue 4 · Mon, Oct 12

   I've read through this week's run. You have a strong set on perinatal mental health, postpartum physiology, and menstrual epidemiology. Two pieces stand out as most worth your time: the equity-centered critique of AI for postpartum depression prevention, which names the real gaps we keep building systems over, and the retinal imaging study on hypertensive disorders in pregnancy, which shows interpretable AI working on a hard obstetric problem with solid external validation. Watch the preprints though, especially the childbirth-related PTSD screener and the menstrual bleeding life-course analysis, both unreviewed and worth flagging to readers as preliminary. No sources failed this week.

4. Issue 3 · Wed, Oct 7

   I found a strong week. The machine learning endometriosis model and the IVF live birth prediction study are both the kind of careful multicenter validation work worth following, though neither breaks new ground clinically yet. The postpartum weight loss RCT (Supporting MumS) looks solid on its face but is still preprint, so hold judgment until peer review. The postpartum depression biomarker paper is intriguing but early, with that remarkably high accuracy figure you'll want to read skeptically. For breadth, the evidence summary on digital tools for postpartum depression screening gives us a clean synthesis to build from. Nothing failed this week, but four items are preprint only, so flag those as provisional.

5. Issue 3 · Tue, Oct 6

   I hit a rate limit on arXiv this week, so the run is incomplete, I missed checking several preprints against the full record. That said, you have solid material: the lactation claims-data phenotype and the environmental chemicals work on ovarian reserve both tackle real gaps in how women's health data gets recorded and what we actually measure. The pelvic pain forecasting study is the kind of practical digital tool that matters for underserved populations. I'd flag that three items are still preprints, so hold them lightly until peer review, and the HCI menopause paper has no abstract to evaluate, you may want to track that one down directly if the title intrigues you.

6. Issue 3 · Mon, Oct 5

   I have a strong week. The endometriosis genetics and methylation work both hit relevance 5, though both are preprints not yet peer reviewed, so we'll want to flag that. The machine learning papers on miscarriage prediction and postpartum depression risk have solid cohort sizes and external validation in at least one case. I'd lead with the iron homeostasis-endometriosis GWAS as the week's anchor, since it actually tackles mechanism rather than just prediction, but pair it with the methylation study so readers see both the statistical promise and the biology questions still open. The maternal mortality excess estimate is important for context but preprint status means we should note we're missing peer review. One HCI piece on menstrual health ecosystems came through without an abstract, so I can't speak to methods or findings there.

7. Issue 2 · Mon, Sep 28

   I found a strong randomized trial on reproductive health messaging for veterans that's worth your time, along with several solid qualitative and scoping review pieces on digital tools for menstrual health, endometriosis, and perinatal care. Three preprints landed this week (abnormal uterine bleeding transcriptomics, the pelvic pain exercise trial, and Ghana maternal health access), so I'd flag those as not yet peer reviewed. The HCI conference papers on menstrual design, menopause in Taiwan, pregnancy loss communities, and menopause cybersecurity all lack abstracts, which means you'll need the full texts to assess them properly. No obvious gaps this run, but the preprints mean the week is still provisional.

8. Issue 1 · Sun, Sep 27

   I found a strong week across AI diagnostics and digital health in women's reproductive medicine. The standout is the scoping review on AI for reproductive health (116 studies, 2021-2025): it's the most comprehensive map we have of this space, and its honesty about bias and validation gaps is exactly what clinicians need to hear before adopting these tools. The plasma proteomics study on adnexal masses (ADAPT-MS) is elegant work, though it's still preprint. Two other preprints worth flagging, the fetal ultrasound framework and the ECG-hormone signatures study, show promise but need peer review before we cite them to readers. On the digital health side, the RCT of the IPT-informed perinatal app is candid: high uptake but no efficacy edge over usual care at 8 weeks, which is the kind of honest result that matters. The menstrual tracking app study raises a real signal about cyberchondria that deserves attention. Three preprints (the antenatal care qualitative work, the French clinician survey, and the tobacco cessation protocol) are early-stage but worth watching.

9. Issue 1 · Sun, Sep 27

   I found a strong week. The RISKmilk framework stands out as a direct answer to a real problem: we have almost no pharmacokinetic data on drugs in breast milk because lactating women were systematically excluded from trials. This AI tool uses molecular structure to predict drug transfer, was validated against clinical data, and is already applied to marketed drugs, that's the kind of practical evidence gap work that matters. The multi-omics infertility study is solid too, integrating vaginal microbiome and inflammatory proteomics in 205 women with reasonable predictive performance and interpretability. I should flag that three pieces are preprints without peer review yet: the MHT-cognition study, the pregnancy prediction model systematic review, and the perimenopause symptom scale. The scale work is interesting for digital phenotyping but early. The IVF AI opinion and the maternal mHealth perspective are both thoughtful but not data-driven studies, so they're more context than evidence. Nothing failed this week, but you'll want to note the preprint status in your framing.