Details

Biopsihosocialni model zdravljenja nosečnic z nosečnostno sladkorno boleznijo
ID Munda, Ana (Author), ID Pongrac Barlovič, Draženka (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (4,28 MB)
MD5: A71AAB63A50411191B6E4EB7D2DD4E3B

Abstract
Uvod: Uniformno zdravljenje nosečnostne sladkorne bolezni (NSB) z naraščajočo incidenco obremenjuje nosečnice in zdravstveni sistem. Želeli smo preveriti, ali se psihosocialni dejavniki povezujejo z uspešnostjo zdravljenja ter oblikovati biopsihosocialni model za stratifikacijo žensk glede na raven tveganja. Metoda: Manj uspešno zdravljenje smo definirali s: (i) slabše urejeno glikemijo (UG), opredeljeno z deležem meritev koncentracije glukoze v krvi v cilju pod 80 % časa in (ii) pojavom neugodnih perinatalnih izidov (NPI), opredeljenim s kompozitom z NSB povezanimi zapleti: rojstvo otroka prevelikega za gestacijsko starost, neonatalna hipoglikemija, zlatenica, zlom ključnice, mrtvorojenost, neonatalna smrt. Z multiplo logistično regresijo smo preverjali povezanost spremenljivk z izidoma. Za oblikovanje napovednih modelov smo uporabili metode strojnega učenja: logistično regresijo, naključni gozd, metodo podpornih vektorjev, XGBoost. Rezultati: Pri vključenih nosečnicah (UG: n = 470, NPI: n = 477; v obeh vzorcih starost = 31,0 [28,0–35,0] kg, ITM = 24,7 [21,6–28,7] kg/m2) so se psihosocialne spremenljivke pomembno povezovale z UG (psihološke: p = 0,005; socialne: p = 0,001) in NPI (psihološke: p < 0,001; socialne: p = 0,028). Biopsihosocialni model je dosegel najvišjo napovedno točnost, še posebej z metodo XGBoost (UG / NPI: PR AUC = 0,87 / 0,81; konkordanca c = 0,93 / 0,91; Brierjev dosežek = 0,11 / 0,14). V modelu UG je pomembnost spremenljivk strmo upadala. V ospredju, z relativno pomembnostjo nad 0,5, so bile spremenljivke ITM matere, teden postavitve diagnoze in opolnomočenost za vodenje NSB. Nasprotno pa je bil upad pomembnosti spremenljivk pri NPI bolj postopen. Relativno pomembnost 0,9 so presegle opolnomočenost za vodenje NSB, teden postavitve diagnoze, starost, vpliv NSB na življenje, samoučinkovitost. Zaključek: Z uspešnostjo zdravljenja se povezujejo psihološke in socialne spremenljivke. Stratifikacija ženske glede na raven tveganja je možna, vendar lahko visoko točnost klasifikacije dosežemo le s kombinacijo biopsihosocialnih spremenljivk. V zaključku dela predstavljamo nadaljnje korake v smeri personalizirane obravnave.

Language:Slovenian
Keywords:nosečnostna sladkorna bolezen, biopsihosocialni model, psihosocialni dejavniki, urejenost glikemije, neugodni perinatalni izidi, strojno učenje, personalizirana obravnava
Work type:Doctoral dissertation
Organization:MF - Faculty of Medicine
Year:2025
PID:20.500.12556/RUL-169799 This link opens in a new window
Publication date in RUL:12.06.2025
Views:832
Downloads:329
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Biopsychosocial model of treatment of pregnant women with gestational diabetes
Abstract:
Background: Uniform gestational diabetes (GDM) treatment with its increasing incidence places a significant burden on healthcare systems and pregnant women. We examined whether psychosocial factors are associated with clinical outcomes and developed a biopsychosocial model for stratifying women by their risk level. Methods: Less successful treatment was defined as (i) poor glycemic control (GC), defined by the proportion of blood glucose measurements in the target being less than 80 % of the time, (ii) the occurrence of adverse perinatal outcomes (APOs), defined as a composite of GDM-related complications, including large-for-gestational age, neonatal hypoglycemia, jaundice, clavicle fracture, stillbirth, neonatal death. We used multiple logistic regression to examine the association with outcomes. Predictive models were developed using machine learning methods: logistic regression, random forest, support vector machine, XGBoost. Results: Among the women included (GC: n = 470, APO: n = 477; in both samples, age = 31,0 [28,0–35,0] years, BMI = 24,7 [21,6–28,7] kg/m2, psychosocial variables were significantly associated with GC (psychological: p < 0,001; social: p = 0,028) and APOs (psychological: p < 0,001; social: p = 0,028). The biopsychosocial model achieved the highest predictive accuracy, especially using XGBoost (GC/APO: PR AUC = 0,87 / 0,81; concordance c = 0,93 / 0,91; Brier score = 0,11 / 0,14). In the GC model, variable importance decreased steeply. The most significant predictors with relative importance above 0,5 were maternal BMI, gestational age at diagnosis, empowerment for GDM management. For APOs, the decrease was more gradual. The relative importance of 0,9 exceeded empowerment for GDM, gestational age at diagnosis, age, the impact of GDM on life, self-efficacy. Conclusion: Psychological and social variables are associated with clinical outcomes. A combination of biopsychosocial variables most accurately predicts women at higher risk for poorer GDM outcomes. We outline the next steps toward personalized care.

Keywords:gestational diabetes, biopsychosocial model, psychosocial factors, glycemic control, adverse perinatal outcomes, machine learning, personalized care

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back