Week of October 31, 2022. Boo.
Last week, a new office patient with a history of Non-Hodgkins Lymphoma recounted how thankful she was to receive a bone marrow transplant (BMT) in 2016. BMT is a tough procedure with many potential complications. It took my patient more than twelve months to feel back to a semblance of normal. Six years out, she is doing well. But her German bone marrow donor's immune system came with a shellfish allergy. If Maryland had a state religion, the crab in all its uses (soups, cakes, etc.) would be the sacrament. Trade-offs permeate healthcare. I can think of no more ironic and Maryland-specific Faustian Bargain than being given a crab allergy for years of life on the Chesapeake Bay.
The patient permitted me to share her story, which is not unique. BMT-acquired food allergies are poorly understood but well documented.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8550919/
and
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5716392/
---
In the U.S., we are at a nadir of reported cases, hospitalizations, and deaths. Anecdotally, I am still hearing about many unreported (and thus uncounted) COVID cases and numerous individuals diagnosed with influenza and RSV identified after days of persistent symptoms and negative home COVID tests.
N.Y. Times Tracker
https://www.nytimes.com/interactive/2021/us/covid-cases.html
Financial Times Data
https://ig.ft.com/coronavirus-chart/?areas=eur&areas=usa&areas=e92000001&areasRegional=usny&areasRegional=usnm&areasRegional=uspr&areasRegional=usaz&areasRegional=usfl&areasRegional=usnd&cumulative=0&logScale=0&per100K=1&startDate=2021-09-01&values=deaths
------
Several summary articles highlighted data on the ineffectiveness of much-touted COVID therapies. Time and more data have demonstrated Hydroxychloroquine, ivermectin, fluvoxamine, and metformin are NOT effective at preventing or mitigating outcomes related to COVID infections. The best strategies are vaccination and masking in crowded or poorly ventilated areas. https://www.nejm.org/doi/full/10.1056/NEJMoa2201662
and
https://www.medscape.com/viewarticle/982852
I have found some interesting discussions about how the evolving coronavirus variants impact the sensitivity and specificity of home rapid antigen test kits. More false negatives mean you need 2-3 tests over 3-4 days at the onset of symptoms. Even then, signs, symptoms, and exposure history are essential to guide behaviors, even in the setting of negative tests.
https://twitter.com/Billius27/status/1580978934531096576
Here is a meta-analysis from spring 2022 that looks at just signs and symptoms as diagnostic tools. Again, the combination of exposure history, local prevalence, symptoms, and testing offer the best chance at diagnosis.
https://pubmed.ncbi.nlm.nih.gov/35593186/
Using non-lab data to diagnose COVID is a fascinating idea. A small study of 31 patients in whom "the first sign of COVID-19 was detected on average nine days before symptoms were reported" via changes in the patient's pulse measured on wrist-worn devices (like FitBits or Apple Watches). A rise in your average heart rate is sensitive for detecting change but not specific for what caused the change. However, when combined with exposure history, symptoms, or other data may offer more specificity as well.
https://www.medrxiv.org/content/10.1101/2022.10.27.22281632v1
Rapid antigen tests are good surrogates for ongoing viral shedding. In other words, for someone who is COVID-positive, using a negative test as the basis for leaving isolation is better than some arbitrary number of days (like 5).
https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2797450
Similar to hurricane tracking, I found a COVID forecasting hub. It compares different forward-looking projection models from various institutions. The models predict cases, hospitalizations, and deaths 10-14 days into the future.
https://viz.covid19forecasthub.org/
Adjacent aside - If you want to learn about forecasting, Yale's School of the Environment's journal article discusses the accuracy of various hurricane tracking models (NOAA/NHC is the winner).
https://yaleclimateconnections.org/2022/06/the-most-reliable-hurricane-models-based-on-their-2021-performance/
Medical Trends and Technology
In late 2021, a study in Nature Aging indicated that users of Viagra and Cialis may have decreased rates of Alzheimer's disease. Expanding the indications for this class of medications is a potential grand slam of therapeutics.
https://www.nature.com/articles/s43587-021-00138-z.epdf
Twitter was very enthusiastic about such data.
https://twitter.com/whippletom/status/1467906736019419139
However, more recent data demonstrated that these drugs are not as valuable for dementia as initially hoped.
https://academic.oup.com/braincomms/article/4/5/fcac247/6731685
This set of articles is a great example of how scientific understanding changes over time. Larger and more granular data sets offer nuance and control for more variables.
Twitter is nearly silent on the new data.
https://twitter.com/search?q=https%3A%2F%2Fdoi.org%2F10.1093%2Fbraincomms%2Ffcac247&src=typed_query
Commentary on both studies
https://www.medscape.com/viewarticle/983094
Infographics
The Amazon/HBO high-budget fantasy T.V. series has ended for the season. Chartr offered a quantitative comparison of Rings vs. Dragons:
https://mcusercontent.com/05dafa9b4317774547b114632/images/37e3086c-32d6-eb61-da8a-ecf5f853aefc.jpg
from
https://read.chartr.co/newsletters/2022/10/28/misery-loves-company
Topics from the newsletter through the "eyes" of our A.I. Overlords!
(What is this section? - https://openai.com/dall-e-2/)
"Painting of a wasp and a dragon holding drinks on the beach"
https://labs.openai.com/s/emFX1Lh7XWycbwo8Y3KzyTXv
and
"oil painting of a wasp, dragon, and crab drinking on the beach wearing fedoras"
https://labs.openai.com/s/6seNA77Y8VLI6FWNnvI1YNud
Things I learned this week
In the spirit of creepy things, I found myself down the "are there really dead wasps in figs?" rabbit hole. (Related - if there are dead wasps in figs, can vegans eat figs?). Spoiler - I see no reason to stop eating figs.
https://www.treehugger.com/are-there-really-wasps-your-figs-4868822
and
https://www.huffpost.com/entry/what-are-figs_n_57bc3dc5e4b03d51368a989a
A.I. visual pattern recognition techniques underly how, for instance, your Ring doorbell camera distinguishes between random movements, a person, an animal, and a package, or your phone camera can identify a subject's face. I learned that there is a robust amount of work in image recognition camouflage - or exploiting the flaws in how imaging algorithms work to confuse an imaging system with atypical patterns or objects. Like COVID tests, A.I. visual tools designed to detect shapes or patterns have some specificity and sensitivity. And, your ugly Christmas sweater may be a useful tool in confusing the killer robots sent from the future to find our mother.
Background
https://viso.ai/applications/computer-vision-applications-in-surveillance-and-security/
Google's inception algorithm sees a turtle as a gun
https://hackaday.com/2017/11/03/googles-inception-thinks-this-turtle-is-a-gun/
Wearing a colored photo on your torso renders humans unrecognizable to the YoLo(v2) algorithm
https://hackaday.com/2019/05/03/the-cloak-of-invisibility-against-image-recognition/
and
https://www.cs.umd.edu/~tomg/projects/invisible/
and
I note the TSA has new 3D CT X-ray scanners (with object identification) at Denver's airport. I wonder how they do with ugly sweaters.
https://www.tsa.gov/news/press/releases/2022/03/18/tsa-awards-7812-million-procure-additional-ct-x-ray-scanners-airport
Clean hands and sharp minds,
Adam
Comments
Post a Comment