PNPs showed previously unknown perineurial hyperplasia and fibrotic dispersion and this was most pronounced in immune-mediated PNPs.
PNPs showed previously unknown perineurial hyperplasia and fibrotic dispersion and this was most pronounced in immune-mediated PNPs.
Explore: pns-atlas.mzhlab.com
A tweetorial 1/13
doi.org/10.1038/s414...
Explore: pns-atlas.mzhlab.com
A tweetorial 1/13
doi.org/10.1038/s414...
#Dementia patients in the neurodegenerative cluster showed more severe disease progression. CSF cell analysis may thus help predict the progression of dementia.
#Dementia patients in the neurodegenerative cluster showed more severe disease progression. CSF cell analysis may thus help predict the progression of dementia.
Interestingly, #MS patients were more likely to have a #progressiveMS when assigned to the neurodegenerative cluster.
Interestingly, #MS patients were more likely to have a #progressiveMS when assigned to the neurodegenerative cluster.
We further used datathin to split our data (with cluster annotations). By training an #XGBoost model on the train set and evaluating on the test, we could validate our clustering. Additionally, we used a second cohort of 3,201 patients to validate our disease clusters.
We further used datathin to split our data (with cluster annotations). By training an #XGBoost model on the train set and evaluating on the test, we could validate our clustering. Additionally, we used a second cohort of 3,201 patients to validate our disease clusters.
To validate the number of clusters, we used datathin method, which splits a random variable into an independent training and test set. @tommytang.bsky.social www.jmlr.org/papers/v25/2...
To validate the number of clusters, we used datathin method, which splits a random variable into an independent training and test set. @tommytang.bsky.social www.jmlr.org/papers/v25/2...
Technical site note: SoupX worked great in this context by identifying top features (diseases) per cluster via TF-IDF (much better than Seurat wilcox test) @tommytang.bsky.social
Technical site note: SoupX worked great in this context by identifying top features (diseases) per cluster via TF-IDF (much better than Seurat wilcox test) @tommytang.bsky.social
Classifying all diseases solely based on the CSF/blood parameters in 8,790 patients resulted in clusters of 4 disease categories: healthy, autoimmune, meningoencephalitis, and neurodegenerative.
Classifying all diseases solely based on the CSF/blood parameters in 8,790 patients resulted in clusters of 4 disease categories: healthy, autoimmune, meningoencephalitis, and neurodegenerative.
We next trained an #XGBoost model on 588 somatoform patients and found that the prediction on the remaining 200 patients showed a strong correlation with true biological age (Pearson r = 0.71).
We next trained an #XGBoost model on 588 somatoform patients and found that the prediction on the remaining 200 patients showed a strong correlation with true biological age (Pearson r = 0.71).
Age primarily affected #Tcells in #CSF and blood, which became activated with age. Interestingly, #CD8Tcells increased in #CSF with age but decreased in blood, suggesting that age-related changes are partially compartment-specific.
Age primarily affected #Tcells in #CSF and blood, which became activated with age. Interestingly, #CD8Tcells increased in #CSF with age but decreased in blood, suggesting that age-related changes are partially compartment-specific.
4/16
In a subcohort of 788 somatoform patients, we found that #Tcells in blood were increased in females, while albumin, protein, IgG, IgA, and IgM ratios were increased in males.
4/16
In a subcohort of 788 somatoform patients, we found that #Tcells in blood were increased in females, while albumin, protein, IgG, IgA, and IgM ratios were increased in males.
We used data-driven approaches to analyze paired #CSF and blood #FlowCytometry of 8,790 (discovery) + 3,201 (validation) patients
A #tweetorial 1/16
We used data-driven approaches to analyze paired #CSF and blood #FlowCytometry of 8,790 (discovery) + 3,201 (validation) patients
A #tweetorial 1/16