Movatterモバイル変換

card {development version}

Next Steps

• cosinor() to be expanded upon to include prediction,and integration into thetidymodels approach in theparsnip package
  • Evaluation of plotting functions for cosinor models
  • Confidence interval methods to be improved upon
• Population cosinor analysis to be reworked for correct predictionsand confidence interval estimates

Deprecations

• The circadian-focused features are being deprecated in thisupcoming release. The goal is to position functions in the appropriatepackage, with the keycosinor() functions to move to aseparate package in a future release.

• The longitudinal event functions are being moved to a separatepackage to make maintenance more straightforward.

card v0.1.1

Updates

• cosinor() now has a stable population mean cosinoroption with appropriate confidence intervals

• procedure_codes() has the latest ICD10 codes, as of11/2023, and are included in the package

Next steps

• Adding additional data analysis tools for cardiology data (includingcatheterization and echocardiogram data)

Deprecations

• The circadian-rhythm features have been deprecated and recurrentdata features have been removed

• Thecosinor() functions will be updated to be morecustomizable and more efficient, however will be moving to a separatepackage by v0.2.0

card v0.1.0

Bugs

• cosinor() unable to run on certain models based on yvalues

Features

• cosinor_features() allows for assessing global/specialattributes of multiple component cosinor analysis
• ggcosinor() is now functional for single and multiplecomponent analysis
• Sequential model building can be performed withbuild_sequential_models(), however it is in a list formatand will likely be updated to be more “tidy” in the future
• Confidence interval methods now work for population-mean cosinor,including summary function
• ggpopcosinor() can show the cosinors for individualsacross a population, along with mean and predicted cosinor
• ggcosinor() accepts single models
• print.cosinor() andplot.cosinor()functions added
• cosinor_zero_amplitude() test added, works forindividual cosinor.
• Population-mean cosinor analysis is added.cosinor()now takes the argument of for individuals. The individual cosinormethods generally work, but may not yet be accurate.
• Circadian rhythm analysis has also created an initial family offunctions that will work to simplify the process of analyzing 24-hourdata. Thecirc_compare_groups() helps to summarizecircadian data by an covariate and time. This is visualized usingggcircadian(). Also includes theggforest() tocreate forest plots of odds ratios. This is dependent on thecirc_odds() function to generate odds ratios by time.
• An important regression function, built with thehardhat package fromtidymodels,cosinor() introduced as a new function to allow fordiagnostic analysis of circadian patterns. Although the algorithm iswell known, having an implementation in R allows potential diagnostics.This includes theggcosinorfit() allows for assessingrhythmicity and confidence intervals of amplitude and acrophase ofcosinor model. Basic methods for assessing the model, such asprint,summary,coef, andconfint currently function.
• Recurrent events can now be analyzed using a powerful functioncalledrecur_survival_table(), which allows for redesigninglongitudinal data tables into a model appropriate for analysis. It isbuilt to extend survival analyses. Therecur_summary_table() function allows for reviewing thefindings from recurrent events by category to help understand eventstrata.
• Thecirc_sun() function allows for identifying thesunrise and sunset times based on geographical location. This isintended to couple with thecirc_center() function tocenter a time series around an event, such as sunrise. A vignette hasbeen added to review this data.