Ali H.Zayter

Epidemiologist || Data Scientist ||Research Methodology || Statistical Methods || ML & AI || CPM || GDPR || HIPAA || Forensic Legal Analytics ||SAS||R||Python||

[Ali H.Zayter @AliHZayter]

Areas of Pure Mathematics that Fuel Artificial Intelligence:
◆ Linear Algebra
◆ Calculus
◆ Probability & Statistics
◆ Set theory
◆ Graph theory
◆ Optimization
◆ Information theory
◆ Topology
◆ Combinatorics

[Frank Harrell @f2harrell]

@AliHZayter @ESteyerberg @BenVanCalster Adding harmonic terms is easy - see an example at the end of https://t.co/8nRzGklSQO

[Ali H.Zayter @AliHZayter]

@ESteyerberg @BenVanCalster @f2harrell or seasonal admissions to the hospital. Any reference to the use of trigonometric functions to model variations of absolute risk across cyclic time metrics.
@ESteyerberg
@f2harrell
@BenVanCalster

[Ali H.Zayter @AliHZayter]

Is it possible to have a discordance between weak /moderate calibration while keeping mean calibration (O:E =1 and alpha = 0) fixed? while share same ICI despite opposite calibration verdicts and or markedly different curve shape /tail.

[Ali H.Zayter @AliHZayter]

@ESteyerberg @BenVanCalster @f2harrell I can't see that the use of Quadratic (lp^2)or cubic spline terms rcs(lp) in calibration curve is efficient when the underlying baseline outcome exhibits seasonal risk patterns such as cardiac events peaking in winter, or infectious diseases incidence.

[Ali H.Zayter @AliHZayter]

@BenVanCalster @ESteyerberg @f2harrell Thank you, great, I am testing that with a simulation study.

[Ali H.Zayter @AliHZayter]

@ESteyerberg @BenVanCalster @f2harrell The ICI results alone (a,b) and (c,d) is masking information that E50 and E90 recover.

[Frank Harrell @f2harrell]

New blog article on Statistical Thinking:confidence limits for bootstrap overfitting-corrected predictive performance measures for regression models, useful for strong internal validations including confidence bands for debiased calibration curves: https://t.co/IiqxnRsK8q #rstats

[Elon Musk @elonmusk]

🇺🇸🇺🇸🇺🇸

[Ali H.Zayter @AliHZayter]

Footage by @badr_elhabsi (Oman).
Me: PCA Projection
♦️Real Camel (3D)
♦️Shadow on Sand (2D)
♦️Sunlight direction: Projection axes; which variance is best preserved.
♦️Sun angles: Different PCA Projections.
♦️Shadow distortion:loss in dimensionality reduction @f2harrell

[Ali H.Zayter @AliHZayter]

من يستثمرُ بالعلم، يبني حضارةً خالدة، ويغرسُ في الزمن عقلاً لا يَشِيخْ.
أشكال الإستثمار بالعلم:
♦️التعليم الجيد
♦️البحث العلمي
♦️المنح الدراسية المحلية والدولية
♦️نشر الثقافة العلمية
♦️التكنولوجيا والابتكار

[Ali H.Zayter @AliHZayter]

GDPR: General Data Protection Regulation

[Ali H.Zayter @AliHZayter]

HIPAA: The Health Insurance Portability and Accountability Act.

[Ali H.Zayter @AliHZayter]

@ESteyerberg @f2harrell @BenVanCalster @GSCollins @laure_wynants Austin PC, Tu JV. Automated variable selection methods for logistic regression produced unstable models for predicting acute myocardial infarction mortality. J Clin Epidemiol. 2004.https://t.co/61NFMJfN9r

[Ali H.Zayter @AliHZayter]

I see that a change in the logistic regression order in terms of variables has an impact on the model's coefficients[effect size estimates]; with a backward or forward regression for the same model variables can lead to a different set of predictors being selected (based on AIC).

[Ali H.Zayter @AliHZayter]

Any reference/simulation study on that?
@ESteyerberg
@f2harrell
@BenVanCalster
@GSCollins
@laure_wynants

[Michael Pyrcz🌻 @GeostatsGuy]

To illustrate the difference between least squares estimation and maximum likelihood estimation, I built an interactive #Python dashboard using @matplotlib.

Adjust the mean of a parametric distribution:

Top plot – the mean squared error (MSE) between the parametric CDF and the empirical CDF of the data.

Bottom plot – the likelihood of the observed data under the parametric PDF.

A visual comparison of how squared error and likelihood differ when estimating parameters. I share it on #github @ https://t.co/lNdq4FF4Bl ∀.

[Ali H.Zayter @AliHZayter]

The Epic Story of Maximum Likelihood
https://t.co/M33HTdVZWE

[Valeriy M., PhD, MBA, CQF @predict_addict]

Looking for good intro to econometrics? This book got it covered.

#econometrics

[Ali H.Zayter @AliHZayter]

Clinical Prediction Models:Chapter 8 (Dealing with Missing Values).
●Missing values in outcome ●Quantification of missingness of predictors
●Patterns of missingness
●Imputation of Missing Predictor Values.
@ESteyerberg

[Daily Dose of Data Science @DailyDoseOfDS_]

NN-SVG: Create neural network architecture drawings parametrically!

Export them to SVG files and use them in your work!