Statistična analiza raziskav bioekvivalence v programskem okolju R

Krajnc, Tajda (author), Grabnar, Iztok (mentor), Neves da Silva, Nuno Miguel Elvas (comentor)

R je prosto dostopen programski jezik in programsko okolje, ki se uporablja predvsem za grafično in statistično analizo. Vsa koda je prosto dostopna, kar pomeni, da je poleg validacije funkcionalnosti možna tudi validacija notranjih struktur. V magistrski nalogi smo preučili uporabo programskega okolja R in razpoložljivih paketov v analizi podatkov pridobljenih v bioekvivalenčni študiji. Bioekvivalenca je pomemben koncept v postopku pridobivanja dovoljenja za promet z zdravilom. Omogoča premostitev predkliničnih in kliničnih študij. Omogoča, da lahko pride zdravilo na trg prej in po nižji ceni. Vrednotenje podatkov, pridobljenih v bioekvivalenčni študiji, je potrebno izvesti z validirano programsko opremo. V nalogi smo v R-ju napisali kodo za statistično analizo podatkov pridobljenih v dveh vrstah načrtov bioekvivalenčne študije. Neponovljeni navzkrižni načrt z dvema periodama in dvema sekvencama ter ponovljeni navzkrižni načrt s štirimi periodami in štirimi sekvencami. V kodi smo uporabili R pakete »NonCompart« za neprostorsko analizo, »BE« za statistično analizo podatkov iz študije navzkrižnega načrta in »replicateBE« za statistično analizo podatkov iz študije ponovljenega navzkrižnega načrta. Kodo smo validirali. V postopku validacije smo uporabili javno dostopne podatke, saj smo želeli omogočiti ponovljivost. Javno dostopne podatke smo analizirali z našo kodo in rezultate primerjali s tistimi pridobljenimi z neodvisno validirano programsko opremo. Dodatno smo prikazali uporabo R v analizi bioekvivalence z uporabo naše kode za analizo dveh internih podatkovnih nizov, navzkrižnega prehoda z enim odmerkom in načrtne študije s popolno ponovitvijo. Ugotovili smo, da v obeh študijah testna formulacija ni bila bioekvivalentna referenčni formulaciji. Priložnost za nadgradnjo našega dela bi bila razvoj lastne kode in paketov ter analiza podatkov brez uporabe že obstoječih paketov. Za prihodnje študije bi bila zanimiva tudi validacija programskega okolja R samega, kot orodja za statistično analizo. Torej validacija notranjih struktur, ne le funkcionalnosti.

• Language: Slovenian
• Keywords: bioekvivalenca, programska oprema R, validacija
• Work type: Master's thesis/paper
• Organization: FFA - Faculty of Pharmacy
• Year: 2022
• PID: 20.500.12556/RUL-143467
• Publication date in RUL: 22.12.2022

Secondary language

• Language: English
• Title: Statistical analysis of bioequivalence studies in R software environment

Abstract

R is an open-source programming language and software environment that is predominantly used for graphics and statistical computing. All code is openly accessible which means that validation of internal structures in addition to functionality is possible. In this thesis we examined the use of R software environment and available packages in bioequivalence data analysis. Bioequivalence is an important concept in medicine approval, allowing bridging of preclinical tests and clinical trials, which enables the medicine to reach the market sooner and at lower cost. Evaluation of data obtained in a bioequivalence study must be performed with validated software. We wrote code for statistical analysis of a single dose crossover design and a full replicate design bioequivalence study in R. We used the package »NonCompart« for noncompartmental analysis, »BE« for the statistical analysis of a single dose crossover design and »replicateBE« for the statistical analysis of a full replicate design study. We validated our code by using it to analyze public access data and comparing our results with those produced by independent validated software. There were no differences between the results produced by our code in comparison with those produced by validated software. We further demonstrated the use of R in bioequivalence analysis by using our code to analyze two inhouse datasets, a single dose crossover and a full replicate design study. We found that in both studies the test formulation was not bioequivalent to the reference formulation. An opportunity to build on our work would be to develop our own code and packages and analyze the data without using already existing packages. For future studies performing “white-box” validation of R as software to perform statistical analysis would be interesting. That is to perform validation of its internal structures, not only of the functionality.

• Keywords: R software, bioequivalence, validation