A systematic review of mobile phone applications for assessment of life threat in the intensive care unit

Izhodišča: V enotah za intenzivno nego se uporabljajo različni prognostični modeli za kvantifikacijo življenjske ogroženosti kritično bolnih. Številni prognostični modeli so uporabljeni v različnih mobilnih napravah, ki so postali del zdravstvene oskrbe. V zaključnem delu smo želeli pregledati značilnosti mobilnih aplikacij za napovedovanje življenjske ogroženosti in določiti najprimernejše aplikacije za uporabo v kliničnem okolju. Metoda: Na podlagi iskalnih nizov, vključitvenih in izključitvenih kriterijev ter priporočil PRISMA (ang. Preferred Reporting Items for Systematic Reviews and Meta-Analyses) smo izvedli sistematični pregled mobilnih aplikacij v dveh spletnih trgovinah (Google Play Store in Apple App Store). V končno analizo smo vključili brezplačne mobilne aplikacije v angleškem jeziku za napovedovanje življenjske ogroženosti v enotah intenzivne nege. V končni analizi je bila kakovost mobilnih aplikacij ocenjena s pomočjo lestvice uMARS (ang. Mobile Application Rating Scale: user version) s strani dveh ocenjevalcev, zaposlenih v enoti za intenzivno terapijo. Mobilne aplikacije smo prav tako med seboj primerjali s pomočjo podatkov pacientov, pridobljenih iz brezplačne podatkovne baze MIMIC III (Medical Information Mart for Intensive Care III). Za kvantitativno analizo mobilnih podatkov smo uporabili Microsoft Excel 2016 in IMB SPSS 23.0 ter programski jezik R za pridobivanje podatkov iz zgoraj omenjene baze. Rezultati: Od 2737 mobilnih aplikacij smo identificirali 20 (1 %) mobilnih aplikacij za oceno življenjske ogroženosti v enotah intenzivne nege. Najpogosteje uporabljeni metodi za oceno življenjske ogroženosti sta bili SOFA (n = 12) in APACHE II(n = 12). Povprečna ocena kakovosti mobilnih aplikacij na Androidu po uMARS oceni je znašala 3,83 ± 0,511, pri čemer je minimalna ocena znašala 2,88, maksimalna pa 4,75. Opazna je bila razlika med ocenami mobilnih aplikacij na operacijskem sistemu Android in iOS, vendar pri tem ni bilo statistično pomembne razlike (p = 0,277). Več kot polovica (60 %) vseh mobilnih aplikacij (n = 12) temelji na znanstvenih dokazih. Na podlagi t-testa smo ugotovili, da je prisotna statistično pomembna razlika med mobilnimi aplikacijami, ki temeljijo na znanstvenih dokazih in med ostalimi (p = 0,001). Na podlagi pridobljenih podatkov iz MIMIC III baze smo preizkusili vse prognostične modele SOFA. Ocena umrljivosti se je v vseh mobilnih aplikacijah med seboj ujemala, odstopanja pa so se pojavila v sami interpretaciji umrljivosti. Diskusija: Mobilne aplikacije se zelo hitro posodabljajo in nastajajo vedno nove, zato je pomembno, da so ocenjene s strani strokovnjakov in raziskovalcev. Le tako lahko zdravstvenim delavcem omogočimo vsebinsko in tehnično ustrezne mobilne aplikacije, ki so primerne za uporabo v kliničnem okolju.Introduction: In intensive care units, different prognostic models for the quantification of life threats for critically ill are used. Many prognostic models are used in various mobile devices that have become part of health care. In the final work, we wanted to quantify the characteristics of mobile applications for predicting life threats and to identify potential applications for use in a clinical environment. Method: Based on the search strings, inclusion and exclusion criteria, and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) recommendations we conducted a systematic review of mobile phone apps in two online stores (Google Play Store and Apple App Store). In the final analysis we included free mobile applications in English for predicting life threats in intensive care units. In the final assessment, we evaluated quality of apps using the uMARS (Mobile Application Rating Scale: user version) scale by two reviewers employed at the intensive therapy unit. We also compared mobile applications with patient data from a free MIMIC III database (Medical Information Mart for Intensive Care III). For the quantitative analysis of mobile data, we used Microsoft Excel 2016 and IMB SPSS 23.0 and the programming language R for obtaining data from the above mentioned database. Results: Out of 2737 apps, we identified 20 (1 %) mobile phone apps for the assessment of life threats in intensive care units. The most commonly used methods for assessing the life threat were SOFA (n = 12) and APACHE II (n = 12). The average assessment of the quality of mobile applications on Android was 3,83 ± 0,511, with a minimum score of 2,88 and a maximum of 4,75. There was a difference between the estimates of mobile applications on Android and iOS, but there was no statistically significant difference (p = 0,277). 60% of all mobile applications (n = 12) are based on scientific evidence. Based on the t test, we found that there is a statistically significant difference between mobile applications based on scientific evidence and among others (p = 0,001). On the basis of the obtained data from the MIMIC III database we tested all the prognostic SOFA models. In all mobile applications, mortality assessments coincided with each other, derogations have occurred in interpretation of mortality. Discussion: The mobile applications are frequently updated and there are new applications developed on a daily basis, so it is important that they are evaluated by experts and researchers. Only in this way can we provide healthcare professionals with a substantially and technically appropriate mobile applications that are suitable for use in clinical environment