Digital solutions in paediatric sepsis: current state, challenges, and opportunities to improve care around the world

Abstract

The digitisation of health care is offering the promise of transforming the management of paediatric sepsis, which is a major source of morbidity and mortality in children worldwide. Digital technology is already making an impact in paediatric sepsis, but is almost exclusively benefiting patients in high-resource health-care settings. However, digital tools can be highly scalable and cost-effective, and-with the right planning-have the potential to reduce global health disparities. Novel digital solutions, from wearable devices and mobile apps, to electronic health record-embedded decision support tools, have an unprecedented opportunity to transform paediatric sepsis research and care. In this Series paper, we describe the current state of digital solutions in paediatric sepsis around the world, the advances in digital technology that are enabling the development of novel applications, and the potential effect of advances in artificial intelligence in paediatric sepsis research and clinical care.

Figure 1:. The digital paediatric sepsis journey

AI=artificial intelligence. CDS=clinical decision support. HTE=heterogeneity of treatment effect.

Figure 2:. Level of implementation, complexity, resource-based suitability, and impact of different digital technology solutions in paediatric sepsis

This figure represents an assessment and illustration of the present and future of the field of digital solutions in paediatric sepsis, performed by the authors, and does not represent a quantitative evaluation of the literature. AI=artificial intelligence. CDS=clinical decision support. EHRs=electronic health records. HICs=high-income countries. LICs=low-income countries. LMICs=low-income and middle-income countries. UMICs=upper-middle-income countries.

Figure 3:. Data-driven phenotyping approaches and validation

(A) Data-driven phenotypes of paediatric sepsis have been studied at different biological levels.^,,, These biological levels have been studied using various types of data sources and imply different levels of complexity. (B) Various unsupervised machine learning algorithms can be used to derive phenotypes, which can then be validated in different ways. HCA=hierarchical clustering analysis. LCA=latent class analysis. GBTM=group-based trajectory modelling. DTW=dynamic time warping. SANMF=subgraph-augmented non-negative matrix factorisation.

Figure 4:. Framework for transferability of adult sepsis algorithms to paediatric cohorts

This figure shows initial algorithm development in adult and paediatric cohorts, followed by incorporation of paediatric-specific factors and computational adjustments. We envision that paediatric sepsis algorithms can borrow from the rich knowledge of adult sepsis prediction models derived from large-scale adult cohorts, and those models derived in other paediatric sepsis cohorts. For example, adult cohort models can be adapted to paediatric cohorts with strategies such as transfer learning and domain adaptation, to teach pretrained sepsis algorithms the characteristics of paediatric sepsis. For implementation purposes, transferred models need to be recalibrated on the target paediatric cohort. Likewise, transfer of paediatric sepsis algorithms to adult algorithms is also possible, necessitating recalibration on the target adult cohorts.