Privacy-preserving multimodal multi-omics and digital phenotyping for first-trimester early warning of maternal–fetal disorders: A systematic review and translational roadmap
Wiku Andonotopo, Muhammad Adrianes Bachnas, Wisnu Prabowo, Eric Edwin Yuliantara, Mochammad Besari Adi Pramono, Julian Dewantiningrum, Efendi Lukas, I. Nyoman Hariyasa Sanjaya, Anak Agung Gede Putra Wiradnyana, Anak Agung Ngurah Jaya Kusuma, Ryan Saktika Mulyana, Khanisyah Erza Gumilar, Ernawati Darmawan, Muhammad Ilham Aldika Akbar, Dovy Djanas, Dudy Aldiansyah, Aloysius Suryawan, Ridwan Abdullah Putra, Theresia Monica Rahardjo, Anita Deborah Anwar, Cut Meurah Yeni, Nuswil Bernolian, Harry Kurniawan Gondo, Laksmana Adi Krista Nugraha, Waskita Ekamaheswara Kasumba Andanaputra, Wibisana Andika Krista DharmaABSTRACT
Early pregnancy provides a limited yet influential opportunity to detect the first traces of maternal–fetal disease before it is biologically visible at a clinical threshold or medically evident. However, with the rapid evolution of omics and digital technologies in health, early warning strategies in perinatology remain fragmented and siloed by data modality and are limited by privacy as well as scale-up issues. We performed a Preferred Reporting Items for Systematic Reviews and Meta-analyses 2020–compliant systematic review to assess whether integrative, privacy-preserving application of multi-omics and digital phenotyping offers the promise of meaningful first-trimester risk stratification for major maternal–fetal disorders. We performed a systematic search of top biomedical databases and registers, screened records for inclusion based on prespecified criteria, and assessed methodological quality and risk of bias with standard tools. A total of 48 studies were included, including transcriptomics, proteomics, metabolomics, lipidomics, epigenomics, microbiome profiling and single-cell and spatial approaches as well as the use of wearable-derived physiological data. Similar evidence is emerging in scale across a range of disorders, including preeclampsia, gestational diabetes mellitus, preterm birth and fetal growth restriction: molecular divergence precedes clinical presentation by weeks to months. Invariable early and accessible signals that continually cropped up across studies were those of cell-free RNA or metabolomic signatures, whereas placental and single-cell data offered more mechanistic explanation as opposed to predictive power in isolation. Integration approaches that considered temporal dynamics and biological hierarchy performed better than single-layer or cross-sectional methodologies but risks of overfitting, population bias, and limited external validation were repeated across studies. Based on the compiled results, early warning less about finding “new” biomarkers as it is respectfully and longitudinally integrating heterogeneous signals. We envision a translational roadmap that includes privacy-preserving analytics, uncertainty-aware models, and equity considerations in all steps from discovery to clinical decision-making. In aggregate, the evidence suggests a recalibration of perinatal testing toward whole-system integrative systems biology in the first trimester – qualified by realism regarding extant limitations and guided by clinical responsibility.