Show HN: Using AI to Help Our Doctors with Fertility(raynbowbaby.com) My wife and I have been trying to conceive for years, so I built a project that open-sources our entire fertility journey—test results, surgeries, medication lists, Creighton Model data, and more. With AI analyzing ~100 fertility factors, I've been able to have better conversations with our doctors. I'd love to hear your thoughts or suggestions! |
First of all, thank you for your courage (and your partner) in sharing your journey in such a useful way.
Our hearts are with you. With sadness, we acknowledge the profound loss that comes with miscarriage. Further, I acknowledge and respect your decision as a couple to not pursue certain treatments that are not in line with your faith.
Andreia Trigo and I set up Enhanced Fertility after our own personal experiences. Our research is focussed primarily on the interpretation of biomarkers, validated screening tools and pelvic/transvaginal ultrasound scans for the detection, diagnosis and treatment planning of infertility with medical staff.
I will write much more when I have time but for now, my quick thoughts:
1. “Wow” for open sourcing your data. What a generous and incredible thing to donate. That decision couldn’t have been easy.
2. Useful to share data in CONTEXT (e.g. age, location, relationship status, treatment history implied by surgery for endometriosis, outcomes such as miscarriage, etc.)
3. Visualisation of biomarkers over time to see how they have changed is highly useful. Each lab or device manufacturer will have conditional reference ranges to know what is normal/abnormal.
4. I can give you the English translations for biomarkers if you want as you have an international audience. Lucky for us we are in London and Portugal with bilingual team members.
5. There are some biomarkers that others would find useful e.g AMH. I know serum AMH may not be useful to track in the context of your partner’s diagnosis but it may be worth me asking our clinical team what other biomarkers might be absent from your data but useful in a clinical context for those patients without a diagnosis.
6. If you have had a semen analysis (with or without DNA fragmentation) or for future male patients it would be useful to understand what parameters are tested and what they mean. It’s part of the diagnostic standard for investigations that include a male partner / patient.
7. Your partner’s pelvic and transvaginal ultrasounds (and other imaging) observations are extremely useful. Antral Follicle Count and Follicle Number Per Ovary are vital indicators you have captured.
8. Your medication plan is unique to you. It’s great to share, but it is highly personalised. Depending on the target audience, it may be better to put a note separating past / current medication, outcomes/side effects your partner experienced, what is a prescribed medication with a duration vs a supplement / vitamin.
9. The screening questionnaire is lacking some very important details. I can provide these privately to you as they relate to you both (and others with a similar diagnosis). Additionally, we are happy to share some of the validated screening tools already used in clinics e.g. FertiQol, PCIQ-F, SCREENIVF, etc. or other tools that help understand the emotional/psychological components as well as the clinical history of a patient and their partner.
10. Use of ChatGPT for analysis… this is the big one. As powerful as o1 is for the context window, number of tokens, etc. it is still quite flawed. We are wrestling with this and have no magical answer. The use of foundation models have tremendous power in helping patients understand things, especially the context of biomarkers and risk factors, but sending a payload with identifiable or sensitive information is a major privacy issue. Again, happy to discuss our research / resources privately on this topic as the trial is not in the public domain yet.
Please reach out to us. You are most certainly a friend we just haven’t met yet.
You are NOT alone.
Frank Sullivan CTO, Enhanced Fertility