Humans are already out of the loop in many cases

Jrinne · February 2, 2025, 3:15pm

Example: Why bother with the unreliable human patient to tell you how well they can see?

JAMA (a well-respected medical journal) article:
Estimating Visual Acuity With Spectacle Correction From Fundus Photos Using Artificial Intelligence

From the article:
"Importance: Determining spectacle-corrected visual acuity (VA) is essential [emphasis mine] when managing many ophthalmic diseases. If artificial intelligence (AI) evaluations of macular images estimated this VA from a fundus image, AI might provide spectacle-corrected VA without technician costs [emphasis mine],…"

It’s already routine for a nurse practitioner to send fundus photos to an AI to diagnose diabetic retinopathy—bypassing the doctor, or really anyone, actually looking at the patient’s retina.

I’ve already been replaced in some instances, but "let’s just skip asking the patient about their vision" is a whole new level.

I don’t love everything about AI. But it’s here, and it’s already EXTREME—whether I like it or not.

Navigating the role of AI—whether in medicine, investing, or elsewhere—won’t be easy for P123 or anyone else. The challenge is knowing how to adapt as the landscape shifts beneath us.

ZGWZ · February 2, 2025, 3:37pm

I've always believed that humans are generally dumber than LLM, so when someone says I'm like an AI, I take it as a compliment, though that doesn't mean LLM is smart enough to do much of anything useful.

Jrinne · February 2, 2025, 3:47pm

They are smart. I gave ChatGPT my CME tests for this round of requirements to keep my medical license. It passed 6 out of 7—missing a pass on neuro-ophthalmology by just one question. These weren’t just rote memorization questions; reasoning was required.

JAMA now reports on AI in almost every issue, and nearly every study shows AI outperforming MDs in diagnostic accuracy on real patient samples. Some publication bias? Absolutely. But considering that JAMA's editorial stance tends to be anti-AI, you’d think there’d be more motivation to disprove the idea that "humans are generally dumber than LLMs." So far, it seems like they’re struggling to win that argument in the article section.

But however that argument shakes out— is AI really so much smarter that an algorithm and an insurance company can determine how well you see without even asking you? That’s a whole new level, and worth taking note of, I think.

ZGWZ · February 2, 2025, 4:01pm

The medical community's research standards are almost completely intolerable in quantitative investing community, so this is not surprising to me. For example, it has long been believed that exercise dramatically reduces mortality (whereas sedentary behavior and BMI increase mortality). But the latest less biased meta-analyses support virtually no actual effect - and even if the correlation after incomplete control is indeed causal, it's far from supporting the current common recommendations for lifestyle interventions. From the perspective of the Quant community, the quality and quantity of research on virtually all interventions and medications (as well as the analysis of those empirical results) is completely unacceptable.

The polygenic score research community raises a number of stock selection-like issues, but they study them with little regard for even the division of the training and test sets, and attribute the decay of PGS predictive validity between races solely to force majeure, rather than mainly to overfitting within the sample. In short, what's good enough for medicine is still far from good enough for personal investment.

Even at the intersection of finance and ML outside of stock selection, the way people control for bias is alarming. For example, default prediction related to real estate loans does not even consider chronology when dividing the training and prediction sets, which leads to the fact that one can “successfully” avoid the intense default period of 06-08 with knowledge from after 08. Corresponding fields outside of finance often do not even collect data that can be used for time series cross-validation.

Curtisbaker3 · February 2, 2025, 4:15pm

Yeah its SOO powerful! I used Chatgpt to build an app that searches pubmed for 6000+ terms, and brings all the related studies. I input an entire lifelabs requisitions into it and now it pulls every related blood test that can be done for any combination of medical diseases. This allows me to instantly see how all the diseases are interconnected and which of the 300+ blood tests are related to a disease based on most available evidence. It summarizes each study very nicely and accurately!

I was super amazed how many conditions overlap with the same underlying causes and risk factors.

I've exported 123 diseases that it mapped out with 1600 studies & associated blood tests from 6,000+ authors across the world.

Look at the crazy amount of data that I can get from it.... Here's an example output for a theoretical person with:
Macular Degeneration
Osteoarthritis
Osteoporosis
Heart Disease
Insomnia
Dementia/Alzheimer's Disease
Cataracts
Anxiety

Test:	Conditions	Rationale
Calcium, Serum	Osteoporosis, Osteoarthritis, Insomnia, Heart Disease, Macular Degeneration	➤ Osteoarthritis: Higher serum calcium concentration inversely associates with radiographic knee osteoarthritis prevalence, suggesting protective potential. Serum Calcium Concentration Is Inversely Associated With Radiographic Knee Osteoarthritis: A Cross-Sectional Study - PubMed ➤ Insomnia: 2022 study found low serum calcium disrupted sleep control and circadian rhythm in shift workers. Lower Serum Calcium Levels Associated with Disrupted Sleep and Rest-Activity Rhythm in Shift Workers - PubMed ➤ Heart Disease: A 2024 study found U-shaped link between calcium levels and mortality in diabetic heart failure patients. U-shaped association between serum calcium and in-hospital mortality in diabetes patients with congestive heart failure: a cohort study - PubMed
Ferritin	Anxiety, Insomnia, Heart Disease, Macular Degeneration	➤ Anxiety: 2020 study found higher risk of anxiety in those with low iron (ferritin) levels. Psychiatric disorders risk in patients with iron deficiency anemia and association with iron supplementation medications: a nationwide database analysis \| BMC Psychiatry \| Full Text A 2024 study found that genetic predisposition to low iron (ferritin) was causually related to higher anxiety. https://pubmed.ncbi.nlm.nih.gov/38547239/ ➤ Insomnia: 2021 study found iron deficiency associated with restless leg syndrome and poor sleep quality in rheumatoid arthritis patients. https://pubmed.ncbi.nlm.nih.gov/34779378/ 2015 study found low ferritin levels correlated with higher periodic leg movements in sleep among older adults. https://pubmed.ncbi.nlm.nih.gov/26498245/ Parent-reported sleep problems linked to low ferritin levels in Egyptian children with ADHD in 2013 study. https://pubmed.ncbi.nlm.nih.gov/24377840/ Autism Spectrum Disorders (ASD) 2013 study found lower ferritin levels linked to sleep fragmentation and periodic limb movements. https://pubmed.ncbi.nlm.nih.gov/24053984/ 2009 study found serum ferritin <45 µg/L linked to higher abnormal sleep movement scores in ADHD children. https://pubmed.ncbi.nlm.nih.gov/19205783/ ➤ Heart Disease: A 2024 study found iron deficiency anemia common in heart failure, treatment practices need improvement. https://pubmed.ncbi.nlm.nih.gov/38572652/ A 2024 study found iron deficiency anemia worsens outcomes in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38563807/ A 2023 study found anemia independently worsens prognosis in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/37649390/ 2024 study found low ferritin levels linked to increased heart failure risk in older adults. https://pubmed.ncbi.nlm.nih.gov/38206230/ 2017 study found both low and high ferritin levels linked to higher heart failure risk. https://pubmed.ncbi.nlm.nih.gov/27976478/ ➤ Macular Degeneration: High levels of iron may be harmful to the eye. 2019 study: Oral iron use may raise retinal hemorrhage risk in wet AMD. https://pubmed.ncbi.nlm.nih.gov/30157115/
Magnesium, Serum	Osteoporosis, Osteoarthritis, Cataracts, Dementia/Alzheimer's Disease, Insomnia, Heart Disease	➤ Osteoarthritis: Higher magnesium intake and serum levels reduce high-sensitivity C-reactive protein (an inflammatory marker) in early radiographic knee osteoarthritis patients. Associations of dietary and serum magnesium with serum high-sensitivity C-reactive protein in early radiographic knee osteoarthritis patients - PubMed (2023): Higher magnesium intake from food and supplements reduces recurrent falls and frailty risk in osteoarthritis patients. Association between Elevated Magnesium Intake and Reduced Risk of Recurrent Falls and Frailty in Osteoarthritis: Data from the Osteoarthritis Initiative - PubMed (2019): Increased dietary magnesium intake is linked to improved knee cartilage architecture, suggesting a preventive role in osteoarthritis. The Association between Dietary Magnesium Intake and Magnetic Resonance Parameters for Knee Osteoarthritis - PubMed Low magnesium intake is associated with increased knee pain and worse function in knee osteoarthritis patients. Low magnesium intake is associated with increased knee pain in subjects with radiographic knee osteoarthritis: data from the Osteoarthritis Initiative - PubMed ➤ Cataracts: A 2021 study found lower magnesium levels in those with cataracts. Association of Serum Magnesium with Oxidative Stress in the Pathogenesis of Diabetic Cataract - PubMed ➤ Dementia/Alzheimer's Disease: A 2018 study found that both low and high plasma magnesium concentrations were associated with a high risk of dementia. Plasma Concentrations of Magnesium and Risk of Dementia: A General Population Study of 102 648 Individuals - PubMed ➤ Insomnia: 2024 study found magnesium and melatonin improved sleep quality but not RMR. Double-blinded trial. https://pubmed.ncbi.nlm.nih.gov/38745424/ 2024 study linked magnesium deficiency score to poor sleep quality, especially in sleep apnea cases. Dose-response observed. https://pubmed.ncbi.nlm.nih.gov/38703902/ 2022 CARDIA study found magnesium intake associated with better sleep quality and less short sleep. Longitudinal analysis. https://pubmed.ncbi.nlm.nih.gov/34883514/ 2021 meta-analysis showed magnesium supplementation reduced sleep onset latency in older adults. Low evidence quality. https://pubmed.ncbi.nlm.nih.gov/33865376/ ➤ Heart Disease: A 2024 study found magnesium depletion score associated with congestive heart failure risk. https://pubmed.ncbi.nlm.nih.gov/37198357/ A 2023 study found magnesium sulfate reduces inflammatory cytokines in coronary artery disease patients. https://pubmed.ncbi.nlm.nih.gov/37665448/ A 2023 study found high calcium/magnesium ratio predicts coronary artery disease in diabetics. https://pubmed.ncbi.nlm.nih.gov/36205093/ A 2021 study found low serum magnesium increases risk of heart failure and atrial fibrillation in type 2 diabetes. https://pubmed.ncbi.nlm.nih.gov/34385344/ A 2020 study found low magnesium and high calcium/magnesium ratio increase mortality in coronary artery disease. https://pubmed.ncbi.nlm.nih.gov/31870835/ A 2020 study found higher dietary magnesium reduces fatal coronary heart disease in postmenopausal women. https://pubmed.ncbi.nlm.nih.gov/31829773/
Ascorbic Acid (Vitamin C)	Cataracts, Dementia/Alzheimer's Disease, Insomnia, Heart Disease	➤ Cataracts: A 2011 study found a 39% lower risk of cataracts in those with higher blood levels of vitamin C. https://pubmed.ncbi.nlm.nih.gov/21705085/ ➤ Dementia/Alzheimer's Disease: A 2024 study found that higher levels of serum vitamin C are associated with slower Alzheimer's disease progression. https://pubmed.ncbi.nlm.nih.gov/38892605/ ➤ Insomnia: A 2024 cohort study found higher dietary Vitamin C intake linked to reduced sleep disorders and sleep apnea risk. https://pubmed.ncbi.nlm.nih.gov/39519494/ ➤ Heart Disease: 2019 study found CAD patients had lower vitamin C and antioxidants in saliva and serum. https://pubmed.ncbi.nlm.nih.gov/31817129/ 2019 study found vitamin C deficiency linked to worse outcomes in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/30211815/
Creatinine (Including eGFR)	Osteoporosis, Osteoarthritis, Dementia/Alzheimer's Disease	➤ Osteoarthritis: Low serum creatinine levels associate with greater knee osteoarthritis presence and severity. https://pubmed.ncbi.nlm.nih.gov/39193760/ ➤ Dementia/Alzheimer's Disease: A 2020 study found that kidney disease is a cause of cognitive decline. https://pubmed.ncbi.nlm.nih.gov/33028210/
CBC + Differential (Complete Blood Count)	Osteoporosis, Insomnia, Heart Disease	➤ Insomnia: 2023 study found anemia linked to insomnia and excessive daytime sleepiness in older adults. http://pubmed.ncbi.nlm.nih.gov/36036443/ 2020 meta-analysis found anemia associated with increased insomnia risk in adults. http://pubmed.ncbi.nlm.nih.gov/33725707/ ➤ Heart Disease: A 2024 study found iron deficiency anemia common in heart failure, treatment practices need improvement. https://pubmed.ncbi.nlm.nih.gov/38572652/ A 2024 study found iron deficiency anemia worsens outcomes in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38563807/ A 2023 study found anemia independently worsens prognosis in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/37649390/ A 2024 study found red blood cell distribution width is a marker of heart failure severity. https://pubmed.ncbi.nlm.nih.gov/39042640/ A 2025 study found CBC-derived inflammation markers predict coronary lesion severity in CAD. https://pubmed.ncbi.nlm.nih.gov/37632217/ A 2013 study found high WBC count predicts coronary heart disease risk. https://pubmed.ncbi.nlm.nih.gov/23467296/
Albumin, Quantitative	Osteoporosis, Cataracts, Heart Disease	➤ Cataracts: A 2005 study found a 50% increased risk of cataracts in those with low serum albumin (a marker of protein intake and nutritional status). https://pubmed.ncbi.nlm.nih.gov/15710820/ ➤ Heart Disease: A 2024 study found low albumin levels predict higher mortality in heart failure patients with CRT-D. https://pubmed.ncbi.nlm.nih.gov/39041845/ A 2024 study found low albumin levels predict higher 28-day mortality in NYHA class IV heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38432040/ A 2024 study found low albumin contributes to higher mortality risk in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38243382/
C-Reactive Protein (CRP)	Osteoarthritis, Dementia/Alzheimer's Disease, Insomnia, Heart Disease, Macular Degeneration	➤ Osteoarthritis: Elevated C-reactive protein levels limit knee muscle strength gains over two years in knee osteoarthritis patients. https://pubmed.ncbi.nlm.nih.gov/24928303/ ➤ Dementia/Alzheimer's Disease: A 2024 study found that those with high inflammation are at higher risk of dementia. https://pubmed.ncbi.nlm.nih.gov/38548879/ A 2022 study found higher inflammation levels were associated with higher risk of alzheimer's disease. https://pubmed.ncbi.nlm.nih.gov/35112776/ ➤ Insomnia: 2023 study found insomnia symptoms linked to elevated CRP in young adults. http://pubmed.ncbi.nlm.nih.gov/30358412/ ➤ Heart Disease: A 2024 study found hs-CRP predicts severity of coronary lesions in premature CAD. https://pubmed.ncbi.nlm.nih.gov/39107719/ A 2024 study found higher CRP levels associated with frailty in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38446241/ A 2024 study found CRP-albumin ratio predicts poor outcomes in heart failure. https://pubmed.ncbi.nlm.nih.gov/38401823/ A 2024 study found elevated hs-CRP predicts heart failure hospitalization in patients with fatty liver disease. https://pubmed.ncbi.nlm.nih.gov/38240197/ ➤ Macular Degeneration: 2021 study: Higher CRP linked to choroidal thinning in intermediate AMD. https://pubmed.ncbi.nlm.nih.gov/34609476/ 2020 study: Late AMD associated with moderately increased systemic CRP. https://pubmed.ncbi.nlm.nih.gov/32937187/ 2020 study: Elevated CRP causally increases AMD risk. https://pubmed.ncbi.nlm.nih.gov/31900758/
Iron/TIBC/% Saturation	Anxiety, Osteoarthritis, Dementia/Alzheimer's Disease, Insomnia, Heart Disease	➤ Anxiety: 2020 study found higher risk of anxiety in those with low iron levels. Psychiatric disorders risk in patients with iron deficiency anemia and association with iron supplementation medications: a nationwide database analysis \| BMC Psychiatry \| Full Text A 2024 study found that genetic predisposition to low iron was causually related to higher anxiety. https://pubmed.ncbi.nlm.nih.gov/38547239/ ➤ Osteoarthritis: Serum iron and ferritin levels are positively associated with osteoarthritis incidence, suggesting iron overload as a risk factor. https://pubmed.ncbi.nlm.nih.gov/39470535/ (2022): A U-shaped relationship exists between iron intake and knee osteoarthritis progression, with optimal intake reducing progression risk. https://pubmed.ncbi.nlm.nih.gov/35458236/ ➤ Dementia/Alzheimer's Disease: A 2023 study found that decreased iron in the blood and increased ferritin in cerebrospinal fluid are associated with Alzheimer's disease, suggesting potential implications for iron metabolism in the development of the disease. https://pubmed.ncbi.nlm.nih.gov/37734209/ ➤ Insomnia: 2021 study found iron deficiency associated with restless leg syndrome and poor sleep quality in rheumatoid arthritis patients. https://pubmed.ncbi.nlm.nih.gov/34779378/ 2015 study found low ferritin levels correlated with higher periodic leg movements in sleep among older adults. https://pubmed.ncbi.nlm.nih.gov/26498245/ Parent-reported sleep problems linked to low ferritin levels in Egyptian children with ADHD in 2013 study. https://pubmed.ncbi.nlm.nih.gov/24377840/ Autism Spectrum Disorders (ASD) 2013 study found lower ferritin levels linked to sleep fragmentation and periodic limb movements. https://pubmed.ncbi.nlm.nih.gov/24053984/ 2009 study found serum ferritin <45 µg/L linked to higher abnormal sleep movement scores in ADHD children. https://pubmed.ncbi.nlm.nih.gov/19205783/ ➤ Heart Disease: A 2024 study found iron deficiency anemia common in heart failure, treatment practices need improvement. https://pubmed.ncbi.nlm.nih.gov/38572652/ A 2024 study found iron deficiency anemia worsens outcomes in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38563807/ A 2023 study found anemia independently worsens prognosis in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/37649390/ A 2023 study found iron deficiency remains undertreated in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38833574/ A 2024 study found iron deficiency common in newly diagnosed heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38803153/ A 2012 study found low total iron binding capacity weakly associated with increased coronary artery disease risk. https://pubmed.ncbi.nlm.nih.gov/22652366/ A 2011 study found low iron status may be associated with ischemic heart disease mortality. https://pubmed.ncbi.nlm.nih.gov/21450624/ 2024 study found low transferrin saturation associated with increased heart failure risk in adults over 40. https://pubmed.ncbi.nlm.nih.gov/39280001/ 2024 study emphasizes importance of transferrin saturation in diagnosing iron deficiency in heart failure. https://pubmed.ncbi.nlm.nih.gov/38567517/ 2023 study found low transferrin levels in heart failure linked to anemia and worse prognosis. https://pubmed.ncbi.nlm.nih.gov/37400990/
Alkaline Phosphatase (ALP)	Osteoporosis, Dementia/Alzheimer's Disease, Heart Disease	➤ Dementia/Alzheimer's Disease: A 2021 study found that those with high ALP are at higher risk of cognitive decline. https://pubmed.ncbi.nlm.nih.gov/32363431/ ➤ Heart Disease: A 2018 study found high alkaline phosphatase levels increase risk of CHD, stroke, and mortality. https://pubmed.ncbi.nlm.nih.gov/30478120/
Lipids Panel (Fasting)	Cataracts, Dementia/Alzheimer's Disease, Heart Disease, Macular Degeneration	➤ Cataracts: A 2021 study found that low HDL cholesterol is a risk factor for developing cataracts. https://pubmed.ncbi.nlm.nih.gov/34128595/ A 2012 study found that high LDL cholesterol may increase risk of cataracts. https://pubmed.ncbi.nlm.nih.gov/23155978/ ➤ Dementia/Alzheimer's Disease: A 2022 study found that low HDL cholesterol and high LDL cholesterol was associated with increased risk of developing Alzheimer's disease. https://pubmed.ncbi.nlm.nih.gov/35124646/ A 2019 study found those with the highest HDL levles had less than half the risk of dementia of others. https://pubmed.ncbi.nlm.nih.gov/30659169/ ➤ Heart Disease: A 2023 study found high total cholesterol increases heart failure risk in postmenopausal women. https://pubmed.ncbi.nlm.nih.gov/37402280/ ➤ Macular Degeneration: 2021 study: High midlife cholesterol correlates with increased AMD risk. https://pubmed.ncbi.nlm.nih.gov/33533136/ 2017 study: High HDL cholesterol causally raises advanced AMD risk. https://pubmed.ncbi.nlm.nih.gov/29025108/ 2021 meta-analysis: DHA, EPA intake lower early AMD risk. https://pubmed.ncbi.nlm.nih.gov/33469697/
Vitamin B12	Anxiety, Cataracts, Dementia/Alzheimer's Disease, Insomnia, Heart Disease, Macular Degeneration	➤ Anxiety: 2019 study linked vitamin B12 deficiency-induced hyperhomocysteinemia to higher depression and anxiety risks in Indian population. https://pubmed.ncbi.nlm.nih.gov/30904222/ ➤ Cataracts: Vitamin B12 was shown in a 2014 study to affect levels of homocysteine, which are related to cataract risk. https://pubmed.ncbi.nlm.nih.gov/25406287/ ➤ Dementia/Alzheimer's Disease: A 2017 study found low levels of vitamin B12 were associated with faster cognitive decline. https://pubmed.ncbi.nlm.nih.gov/28513119/ ➤ Insomnia: 2023 study found that low vitamin B12 levels is a risk factor for insomnia. https://pubmed.ncbi.nlm.nih.gov/38063594/ ➤ Heart Disease: 2013 study found high vitamin B12 deficiency linked to cardiovascular risk factors in Indian CAD patients. https://pubmed.ncbi.nlm.nih.gov/23473764/ 2009 study found vitamin B12 deficiency associated with CAD in Indian population, especially vegetarians. https://pubmed.ncbi.nlm.nih.gov/19676146/ 2006 study found vitamin B12 deficiency contributes to premature CAD via hyperhomocysteinemia in men under 45. https://pubmed.ncbi.nlm.nih.gov/17002799/ 2004 study found folic acid with vitamin B12 lowers homocysteine in CAD patients better than vitamin B6. https://pubmed.ncbi.nlm.nih.gov/14985687/ ➤ Macular Degeneration: 2013 study: Elevated homocysteine, low vitamin B12, folate predict AMD risk. https://pubmed.ncbi.nlm.nih.gov/23636242/ 2007 study: Higher homocysteine and low B12 increase AMD risk. https://pubmed.ncbi.nlm.nih.gov/17258528/ 2009 trial: Folic acid, B6, B12 supplementation lowers AMD risk in women. https://pubmed.ncbi.nlm.nih.gov/19237716/
Selenium, Plasma	Anxiety, Osteoarthritis, Cataracts, Heart Disease	➤ Anxiety: 2022 study found lower selenium levels associated with higher anxiety in children. https://pubmed.ncbi.nlm.nih.gov/34686396/ ➤ Osteoarthritis: (2022): Lower plasma selenium levels are associated with higher prevalence of radiographic osteoarthritis. https://pubmed.ncbi.nlm.nih.gov/35166315/ ➤ Cataracts: A 2018 study found an eight times higher risk of cataracts in those with low selenium. https://pubmed.ncbi.nlm.nih.gov/30260193/ ➤ Heart Disease: A 2024 study suggests selenium may benefit heart failure via immunoregulatory mechanisms. https://pubmed.ncbi.nlm.nih.gov/38295576/ A 2024 study found high selenium linked to diabetes, low selenium linked to heart failure. https://pubmed.ncbi.nlm.nih.gov/37996718/ A 2023 study found selenium deficiency increases readmission in new heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38171821/ A 2022 study found high selenium levels reduced mortality and new heart failure in non-smokers. https://pubmed.ncbi.nlm.nih.gov/34931411/ A 2022 study found higher selenium levels reduce mortality in type 2 diabetes patients. https://pubmed.ncbi.nlm.nih.gov/34664061/
Cortisol, AM Serum	Anxiety, Dementia/Alzheimer's Disease, Insomnia	➤ Anxiety: Assesses adrenal function; assesses hormone response to stress ➤ Dementia/Alzheimer's Disease: A 2013 study found that high cortisol may be a risk factor for Alzheimer's disease. https://pubmed.ncbi.nlm.nih.gov/23611893/ ➤ Insomnia: 2024 study found higher cortisol levels linked to poor sleep architecture in menopausal women. http://pubmed.ncbi.nlm.nih.gov/38909441/
DHEAS (Dehydroepiandrosterone Sulphate)	Macular Degeneration	➤ Macular Degeneration: 2007 study: Lower DHEAS levels correlate with severe AMD. https://pubmed.ncbi.nlm.nih.gov/17157799/
Protein (Total)	Osteoporosis
Folate, Serum	Anxiety, Osteoarthritis, Cataracts, Dementia/Alzheimer's Disease, Insomnia, Heart Disease, Macular Degeneration	➤ Anxiety: Low folate may lead to high homocysteine levels. 2019 study linked hyperhomocysteinemia to higher depression and anxiety risks in Indian population. https://pubmed.ncbi.nlm.nih.gov/30904222/ ➤ Osteoarthritis: Higher folate intake is associated with reduced knee pain, improved function, and slower osteoarthritis progression in at-risk groups. https://pubmed.ncbi.nlm.nih.gov/39246020/ Folate deficiency correlates with increased clinical and radiological severity of knee osteoarthritis, emphasizing folate's role in disease progression. https://pubmed.ncbi.nlm.nih.gov/38329818/ ➤ Cataracts: A 2014 study found a 24% increased risk of cataracts in those with lower folate levels. https://pubmed.ncbi.nlm.nih.gov/25406287/ ➤ Dementia/Alzheimer's Disease: A 2022 study found a 68% increased risk of dementia in those with folate deficiency. https://pubmed.ncbi.nlm.nih.gov/35292483/ ➤ Insomnia: 2024 study found an inverted L-shaped association between serum folate levels and sleep duration. https://pubmed.ncbi.nlm.nih.gov/39612384/ ➤ Heart Disease: 2001 study found low folate intake associated with increased acute coronary events in men. https://pubmed.ncbi.nlm.nih.gov/11390336/ 2008 study found low folate linked to CHD independently of homocysteine in Turkish men. https://pubmed.ncbi.nlm.nih.gov/18694616/ ➤ Macular Degeneration: 2013 study: Elevated homocysteine, low vitamin B12, folate predict AMD risk. https://pubmed.ncbi.nlm.nih.gov/23636242/ 2007 study: Higher homocysteine and low B12 increase AMD risk. https://pubmed.ncbi.nlm.nih.gov/17258528/ 2009 trial: Folic acid, B6, B12 supplementation lowers AMD risk in women. https://pubmed.ncbi.nlm.nih.gov/19237716/ 2016 study: Higher folate intake reduces progression to geographic atrophy. https://pubmed.ncbi.nlm.nih.gov/26961928/
Hemoglobin A1C (HbA1C)	Cataracts, Dementia/Alzheimer's Disease, Insomnia	➤ Cataracts: A 2021 study found that high blood sugar is a risk factor for developing cataracts. https://pubmed.ncbi.nlm.nih.gov/34128595/ ➤ Dementia/Alzheimer's Disease: A 2022 study found a 22% increased risk of cognitive decline in those with uncontrolled blood sugar levels. https://pubmed.ncbi.nlm.nih.gov/35742517/ A 2015 study found nearly triple the risk of dementia in those with diabetes. https://pubmed.ncbi.nlm.nih.gov/25524954/ ➤ Insomnia: Low sleep may lead to high blood sugar levels. 2022 study found HbA1c elevation correlated with poor sleep quality and duration in prediabetic UK patients. https://pubmed.ncbi.nlm.nih.gov/35166578/ 2023 study found HbA1c linked to OSA severity, with stronger associations in middle-aged women. https://pubmed.ncbi.nlm.nih.gov/36437403/
Mercury, Blood	Dementia/Alzheimer's Disease, Insomnia	➤ Dementia/Alzheimer's Disease: A 2018 study found a higher level of mercury in those with alzheimer's disease. https://pubmed.ncbi.nlm.nih.gov/29439342/ ➤ Insomnia: 2020 study linked higher mid-childhood mercury levels to delayed sleep timing in adolescents. Sex-specific patterns. https://pubmed.ncbi.nlm.nih.gov/32956656/
Aspartate Transaminase (AST)	Osteoporosis
Zinc, Serum	Anxiety, Insomnia, Insomnia, Heart Disease, Macular Degeneration	➤ Anxiety: 2024 review confirmed lower zinc levels and intake associated with anxiety, supporting zinc supplementation. https://pubmed.ncbi.nlm.nih.gov/37364014/ ➤ Insomnia: 2020 RCT found zinc supplementation improved physical quality of life in premenstrual women, but sleep improvements were non-significant. https://pubmed.ncbi.nlm.nih.gov/32514756/ 2018 RCT found zinc supplementation improved sleep quality, latency, and serum zinc levels in ICU nurses. https://pubmed.ncbi.nlm.nih.gov/29241421/ 2015 cohort study found higher zinc concentrations correlated with better sleep quality in adolescents but not preschoolers. https://pubmed.ncbi.nlm.nih.gov/26184300/ ➤ Insomnia: A 2022 study found serum zinc and zinc/copper ratio were inversely related to long sleep duration in adults. https://pubmed.ncbi.nlm.nih.gov/34453310/ A 2012 study linked serum zinc/copper ratios and sleep duration, suggesting higher ratios improve sleep in adult women. https://pubmed.ncbi.nlm.nih.gov/22476977/ ➤ Heart Disease: 2011 study found low zinc levels and decreased Zn/Cu ratio in severe heart failure patients. https://pubmed.ncbi.nlm.nih.gov/21258970/ ➤ Macular Degeneration: 2013 review: Zinc may help prevent advanced AMD progression. https://pubmed.ncbi.nlm.nih.gov/23652490/ 2008 trial: Zinc-monocysteine improves macular function in dry AMD. https://pubmed.ncbi.nlm.nih.gov/18600492/ 2007 study: Zinc supplementation reduces plasma oxidative stress in AMD. https://pubmed.ncbi.nlm.nih.gov/17157802/ 2001 trial: Antioxidants and zinc reduce AMD progression and vision loss. https://pubmed.ncbi.nlm.nih.gov/11594942/
Thyroid Stimulating Hormone (TSH)	Anxiety, Dementia/Alzheimer's Disease, Insomnia, Heart Disease	➤ Anxiety: 2005 study found that those with imbalanced thyroid had more panic attacks. https://pubmed.ncbi.nlm.nih.gov/15610948/ ➤ Dementia/Alzheimer's Disease: A 2016 study found almost double the risk of dementia in those with hypothyroidism. https://pubmed.ncbi.nlm.nih.gov/27558885/ ➤ Insomnia: 2022 study found isolated elevated TSH associated with poor sleep, with TSH normalizing after improved sleep. https://pubmed.ncbi.nlm.nih.gov/34432233/ ➤ Heart Disease: 2024 study suggests higher TSH may increase unstable angina risk but reduce heart failure risk in euthyroid patients. https://pubmed.ncbi.nlm.nih.gov/38518006/ 2021 study found worsening TSH levels over time linked to poorer outcomes in chronic heart failure patients. https://pubmed.ncbi.nlm.nih.gov/33938411/ 2023 study found normal TSH levels have U-shaped link with CAD severity in nondiabetic patients. https://pubmed.ncbi.nlm.nih.gov/36680358/
Omega-3 Fatty Acid	Osteoarthritis, Insomnia	➤ Osteoarthritis: Omega-3 polyunsaturated fatty acids are inversely associated with osteoarthritis risk in adults, but causality is unconfirmed. https://pubmed.ncbi.nlm.nih.gov/38760818/ ➤ Insomnia: Double-blind 2018 study found omega-3 PUFA improved depression, anxiety, and sleep in outpatients with major depressive disorder. https://pubmed.ncbi.nlm.nih.gov/30317101/ 2018 study found erythrocyte n-3 PUFAs inversely associated with sleep disturbance in Chinese children and adolescents. https://pubmed.ncbi.nlm.nih.gov/29534525/ 2017 study found sleep apnea patients post-PCI had lower EPA/AA ratios, associated with sleep apnea severity. https://pubmed.ncbi.nlm.nih.gov/28631078/
Arsenic, Random Urine	Dementia/Alzheimer's Disease, Insomnia	➤ Dementia/Alzheimer's Disease: A 2020 study found that higher arsenic levels in the environement is associated with worse outcomes in those with alzheimer's disease. https://pubmed.ncbi.nlm.nih.gov/31962196/ ➤ Insomnia: 2017 study found environmental chemicals, including arsenic and heavy metals, linked to sleep disturbances in adults. https://pubmed.ncbi.nlm.nih.gov/27858272/
Chloride	Heart Disease	➤ Heart Disease: A 2024 study found U-shaped link between chloride levels and mortality in heart failure patients. https://pubmed.ncbi.nlm.nih.gov/38556334/ A 2023 study found U-shaped association between chloride and mortality in elderly heart failure. https://pubmed.ncbi.nlm.nih.gov/37349910/ A 2023 study found persistent low chloride worsens outcomes in acute heart failure. https://pubmed.ncbi.nlm.nih.gov/36584943/
Gamma-Glutamyl Transferase (GGT)	Osteoporosis
Thyroxine Free (FT4)	Anxiety, Dementia/Alzheimer's Disease	➤ Anxiety: 2005 study found that those with imbalanced thyroid had more panic attacks. https://pubmed.ncbi.nlm.nih.gov/15610948/ ➤ Dementia/Alzheimer's Disease: A 2016 study found almost double the risk of dementia in those with hypothyroidism. https://pubmed.ncbi.nlm.nih.gov/27558885/
Glucose-6-Phosphate Dehydrogenase (G6PD) including CBC ($33.61 + $8.72)	Cataracts	➤ Cataracts: G6PD is a genetic condition that occurs in 2% of people that may cause cataracts. https://pubmed.ncbi.nlm.nih.gov/15235179/
Sodium	Osteoporosis, Heart Disease	➤ Heart Disease: A 2024 study found high sodium/chloride ratio predicts mortality in acute heart failure. https://pubmed.ncbi.nlm.nih.gov/38570623/
Alanine Transaminase (ALT)	Osteoporosis
Thyroperoxidase Antibody (TPO)	Anxiety	➤ Anxiety: 2005 study found that those with imbalanced thyroid had more panic attacks. https://pubmed.ncbi.nlm.nih.gov/15610948/
Urate (Uric Acid), Serum	Osteoarthritis, Heart Disease	➤ Osteoarthritis: Asymptomatic hyperuricemia is independently associated with knee osteoarthritis https://pubmed.ncbi.nlm.nih.gov/36196710/ Serum uric acid and hyperuricemia are associated with osteophytes in knee osteoarthritis, particularly in females. https://pubmed.ncbi.nlm.nih.gov/26743214/ ➤ Heart Disease: A 2023 study found hyperuricemia is an independent risk factor for coronary heart disease. https://pubmed.ncbi.nlm.nih.gov/37524613/ A 2023 study found hyperuricemia increases mortality risk in chronic heart failure patients. https://pubmed.ncbi.nlm.nih.gov/37091841/ 2022 study found high urate levels increase coronary heart disease risk, especially in men. https://pubmed.ncbi.nlm.nih.gov/35197211/
Homocysteine	Anxiety, Cataracts, Dementia/Alzheimer's Disease, Macular Degeneration	➤ Anxiety: 2019 study linked hyperhomocysteinemia to higher depression and anxiety risks in Indian population. https://pubmed.ncbi.nlm.nih.gov/30904222/ ➤ Cataracts: A 2014 study found a 17% increased risk of cataracts in those with higher homocysteine levels. https://pubmed.ncbi.nlm.nih.gov/25406287/ ➤ Dementia/Alzheimer's Disease: A 2022 study found higher risk of dementia in those with high homocysteine. https://pubmed.ncbi.nlm.nih.gov/36155508/ ➤ Macular Degeneration: 2013 study: Elevated homocysteine, low vitamin B12, folate predict AMD risk. https://pubmed.ncbi.nlm.nih.gov/23636242/ 2007 study: Higher homocysteine and low B12 increase AMD risk. https://pubmed.ncbi.nlm.nih.gov/17258528/ 2009 trial: Folic acid, B6, B12 supplementation lowers AMD risk in women. https://pubmed.ncbi.nlm.nih.gov/19237716/
Estradiol/Estrogen, Serum	Insomnia	➤ Insomnia: A 2019 study showed transdermal estradiol improves sleep metrics independent of vasomotor and depressive symptoms. https://pubmed.ncbi.nlm.nih.gov/31688579/ A 2015 trial found estradiol and venlafaxine improved insomnia and sleep quality in women with hot flashes. https://pubmed.ncbi.nlm.nih.gov/25325454/ A 2011 study linked increasing estradiol and better sleep quality to improved mood during menopause, but not hot flash suppression. https://pubmed.ncbi.nlm.nih.gov/21525161/ 2008 study found sleep variation correlated with estradiol levels in women of reproductive age. https://pubmed.ncbi.nlm.nih.gov/18714190/
Anti-Thyroglobulin	Anxiety	➤ Anxiety: 2005 study found that those with imbalanced thyroid had more panic attacks. https://pubmed.ncbi.nlm.nih.gov/15610948/

Jrinne · February 2, 2025, 4:25pm

Nice!

This is well established. The AREDS and AREDS2 multi-center studies confirmed that a combination of lutein, vitamin C, vitamin E, zinc, and copper is effective in slowing the progression of age-related macular degeneration (AMD).

The copper is included because zinc competes with copper for absorption in the gut, and adding copper helps prevent a deficiency.

No doubt. And well-ingrained into clinical practice.

BTW, the full medical tern for the macula is macula lutea because of its yellow color. Turns out, it’s yellow because the body actively concentrates lutein in the macula.

Age-Related Eye Disease Studies (AREDS/AREDS2)
nih.gov

ZGWZ · February 2, 2025, 4:29pm

Yes, LLM would greatly help medical practitioners as good productivity tool, but as a result it would further increase mortality and decrease health. The reason is simple: not only are the samples in these studies too small, but they are even largely unable to distinguish between obvious inversions of cause and effect. For example, the first paper had only 84 samples and was eager to assume that serum zinc levels were the cause of seborrheic dermatitis without considering the obvious other possibilities that seborrheic dermatitis itself alters serum zinc levels or that the two have a common cause.

The third paper titled meta-analysis also still had only 2,685 patients and like (almost) all medical meta-analyses didn't even use the undercorrected FAT-PET-PEESE publication bias correction to blithely conclude that there was no publication bias. Still, this paper at least doesn't foolishly infer causality and add that more research is needed. But by the standards of the economics profession, the reliance on secret private data and the severe lack of publication bias correction means that this amounts to almost zero: even the correlation itself has no credibility whatsoever.

As a result, LLM's greater ease of access to these studies by practitioners would further degrade the current quality of care.

ZGWZ · February 2, 2025, 4:45pm

Neither their data nor their conclusions of the meta analysis seem to support that zinc interventions are effective.

Honestly, we would need far more invasive data acquisition and sharing than we have now if interventions of this detail could be shown to be effective for specific areas of disease (rather than all-cause mortality). I doubt a system like that would be cost effective though.

Even in this study, which lacked SES controls, zinc intake above 5.6 mg per day still did not help to reduce all-cause mortality in men, let alone the current recommendation of 11 mg. Since zinc intake tends to rely on foods that are relatively expensive per unit (e.g., oysters, beef), this correlation is still likely the result of differences in SES captured by this variable.

However, this study has the benefit of having a much larger sample than NHANES and AREDS, and the outcome/endpoint included all-cause mortality. This is because it appears that zinc intake decreases CVD mortality and cancer mortality but increases other mortality rates. As a result, we can know the optimistic analysis of the others is very likely to be more biased

Jrinne · February 2, 2025, 4:55pm

Here’s a news release about a more recent re-evaluation of the AREDS2 study for geographic atrophy (GA):

Supplements slow disease progression during late stage of “dry” age-related macular degeneration

While the article doesn’t explicitly state it, this puts the AREDS2 over-the-counter vitamin combination in a position where it might be superior to the more expensive—and, for some, more cumbersome—injections of medications directly into the eye.

SYFOVRE (pegcetacoplan) and IZERVAY (avacincaptad pegol) were only recently FDA approved for treatment of geographic atrophy and require injections into the eye. Their results for these medications were pretty bleak but statistically significant.

AREDS2 is already the standard of care and Medicare actively monitors physician compliance with its use in electronic health records. Unlike recent AI-driven advancements in ophthalmology, this is a case where a simple, well-researched nutritional approach appears to outperform high-tech, high-cost pharmaceutical interventions.

Most of this pre-dates widespread use of LLMs.

ZGWZ · February 2, 2025, 5:11pm

Indeed, nutritional supplements are at least less costly and less harmful in their effects than more costly and invasive high-tech interventions.

Curtisbaker3 · February 3, 2025, 5:01pm

Hmm those are some good points there. Thanks for the feedback.

Sounds like the next steps would be to test if it can:

Rank studies by sample sizes
Prioritize studies that better show cause and effect (bidirectional mendelian studies, RCTs, etc)
Screen out studies that don't adjust for confounding (common cause).
Screen for common publishing biases

In the Korean study you posted - it looks like they did find that low zinc intake was associated with higher mortality rates? Or on the flip side - higher zinc is lower mortality compared to low zinc?

"lower zinc intake was significantly associated with all-cause and CVD mortality (≤5.60 mg/day vs. >7.98 mg/day; HR, 1.13; 95% CI, 1.01–1.25 for all-cause mortality; and ≤5.12 mg/day vs. >7.28 mg/day; HR, 1.42; 95% CI, 1.11–1.81 for CVD mortality) but not cancer mortality (≤5.60 mg/day vs. >10.08 mg/day; HR, 1.09; 95% CI, 0.90–1.33 ). Similar trends were observed in men and women."

@Jrinne - that's pretty cool! Didn't realize the macula lutea is yellow because of the lutein in the macula!

ZGWZ · February 4, 2025, 12:22am

First, the threshold of 5.6 mg/day is well below the traditionally recommended intake of zinc

Second, there may still be confounding from uncontrolled confounders such as varying severity of pre-existing conditions and SES, so data from the subgroup of disease-free never-smokers and controlling for SES are needed, but there may still be uncontrolled untested pre-existing conditions

Finally, there may not be any effect. Still, the effect may not be positive and only a sufficiently large, low-bias and well-designed RCT would be able to detect this difference adequately. This is costly and may not rationalize its fixed costs.

Edit:

Since the difference in all-cause mortality between the highest and lowest zinc intake groups is much larger in the unadjusted model (1.8) than in the final model with grossly incomplete adjustment (1.13), it is tempting to assume that further, more refined adjustments are very likely to eliminate, or even reverse, the ratio of all-cause mortality between the two.

“Mendelian randomization” is not any ‘randomization’, but only correlations after control for ‘genetic scores’ that are themselves highly biased. But even if the "polygenic score" controls are complete and precise, they still grossly ignore differences in SES and pre-existing diseases. A better illustration of causal effects is the sibling study. This is because a large proportion of unobservable socio-economic background factors and genetic factors can be controlled for with family fixed effects.

And RCTs still have serious publication bias problems and have smaller examples because of costs (maybe), especially when there is bias at all three levels of methodology/researcher/meta-analyzer, as there is in the case of Mindfulness Therapy (or related other therapies like ACT/ Hakomi/Morita/DBT/MDT), which is serious even by the standards of the medical profession.

Even if you do all this meta-analytic work, you'll quickly find that the available samples are too small to analyze and/or adjust for bias without getting encouraging results. And it is very easy for publication bias adjustments to be incomplete because even the current state-of-the-art FAT-PET-PEESE significantly overestimates the effect size of the market capitalization factor - a technique under which almost every major current finding in economics has been shown to be either insignificant or to have a significant, but almost meaningless, effect size! The latter, in turn, are often easily proven to be still overestimated in further research by expertise from specialized fields as the example mentioned above

By the standards of factor investing, the quantity and quality of data currently available to the medical community is too low to produce reliable results. It is virtually impossible to obtain data in the medical community that is similar, or even simply equivalent, to the quantity and quality of CRSP, yet necessary to demonstrate causality, not just correlation, as required by stock market investors.

Edit2:

Medicine has the advantage of less time-varying factor returns than the stock market, and the disadvantage is that it is far more difficult to estimate “stock prices” and “returns”. We can only rely on relatively precise (but still very imprecise) endpoint of death or other endpoints (such as the diagnosis of a particular disease) that are common but problematic to judge the effectiveness of an intervention. And, while the costs of medical interventions are relatively easy to estimate compared to the difficulty of calculating transaction costs in the stock market, the difficulty lies in the calculation of returns. For stock investments, a dollar is a dollar, whereas it is much harder to measure the economic benefits of interventions in medicine.