Maternal and child health (MCH) outcomes are good indicators of whether a health system works, and for whom. In rich countries, preventable maternal deaths are concentrated in poor neighbourhoods and marginalised populations. In low- and middle-income countries (LMICs) they reflect deeper and broader system failures that overwhelm clinical services.
In both contexts, digital twins offer a new way to understand maternal and child health outcomes as the output of complex, dynamic systems and a potential means of testing interventions and acting earlier and more precisely.
Image: OpenAI
A brief history: before “digital twins”
In the 1990s, David Eddy and colleagues developed Archimedes, a large-scale, equation-based simulation of human physiology, disease progression and healthcare delivery. Archimedes could model individual patients moving through care pathways and aggregate them into virtual populations to test clinical guidelines and policy choices.
Archimedes was remarkably innovative but ahead of its time. It lacked real-time data, and was slow and unsuited to operational use for day-to-day decision-making. Modern digital twins address many of these limitations. Built not only on health and biological data but social, environmental, and system-level information, they can be used for health system planning, learning and action.
What is a population-level digital twin?
Digital twins can simulate a mother, foetus or infant, and their biology. A population-level digital twin is a continuously updated virtual representation of a higher-level system such as a clinical pathway, a hospital or a community. This kind of digital twin integrates multiple data streams:
- routine clinical data (EHRs, claims)
- public health data (births, deaths, immunisation)
- social determinants (housing, income, transport, food access)
- environmental data (air quality, heat, water)
- system variables (staffing, service availability, referral delays).
The twin updates over time and can be used to ask questions such as:
- What would happen if we expanded antenatal care in this district?
- What if community health workers were added here, but not there?
- Which combination of interventions reduces preterm birth, and for whom?
Eliminating health disparities in rich countries
The US spends more on healthcare than any other country yet maternal mortality is rising, with stark racial disparities. Black women are roughly three times more likely to die from pregnancy-related causes than white women. Infant mortality follows the same geographic and socioeconomic pattern.
In this setting, digital twins could help address two persistent limitations of existing MCH programmes: they tend to be clinically narrow (focused on guidelines, checklists, or single conditions, without addressing access, continuity or social context) and population-blind (applied uniformly across regions and groups, assuming that what works on average will work everywhere).
In Northeast Ohio, for example, population-level digital twins are being used to model neighbourhood-level health disparities by integrating clinical outcomes, socioeconomic conditions, environmental exposures and service availability.
This allows regional health systems and public agencies to identify and understand where and why maternal risk is concentrated, and test which combinations of interventions would reduce risk.
The models generally show that clinical care alone is insufficient. For example, expanding high-risk obstetric services improves outcomes only if transport barriers are addressed. Better maternity facilities help only if staffing and referral pathways are fixed. Prenatal screening works only if follow-up care is accessible and trusted.
Digital twins help make these interdependencies visible, allowing more effective redesign of care pathways and allocation of resources.
Improving maternal and child health in LMICs
In LMICs, systemic maternal and child health challenges include fragmented care, overburdened staff and weak referral systems resulting in delayed care.
The result is mortality and severe morbidity at much higher rates:
- In 2023, the maternal mortality ratio in low-income countries was more than 30 times higher than in high-income countries, with a 1 in 65 lifetime risk of maternal death versus approximately 1 in 8,000 in high-income settings.
- Neonatal and stillbirth rates in low-income countries are ten times or more higher.
- Infant mortality rates in low-income countries are often below 5 deaths per 1,000 live births; in poorer countries they can exceed 50 per 1,000, a ten-fold difference.
It is tempting on this basis to portray LMIC health systems as lagging behind high-income countries. This is somewhat misleading, as many deliver reasonably good care with dramatically fewer resources by relying for example on primary care and community health workers rather than specialists, integrating well, and using population-level thinking. Lessons for the rich world?
Digital twins as policy level decision-support
In LMIC settings, the value of digital twins is less in real-time clinical micromanagement and more in strategic learning at lower risk and cost.
For example, a district-level maternal health digital twin can simulate:
- the impact of adding community health workers versus expanding clinics
- the effect of improving transport versus adding ultrasound capacity
- how nutrition support interacts with antenatal care attendance.
This allows governments and NGOs to prioritise interventions, avoid costly mistakes and adapt programmes to local conditions. The learning happens in silico instead of trial-and-error methods on vulnerable populations.
The South African use case
South Africa is a middle-income economy with high levels of inequality and deep public-sector constraints. It has relatively strong academic health centres. This may make it an ideal environment for population-level digital twins in MCH
Potential applications include:
- modelling referral delays between clinics and district hospitals
- understanding where obstetric emergencies cluster geographically
- simulating the impact of preoperative clinics for C-sections
- testing hybrid care models using community health workers plus digital support.
Digital twins are only as good as their underlying data. Population-level digital twins need datasets that are internally consistent, linked across levels of the system, and stable enough over time to preserve relationships and trends. Partial or imperfect data can still be informative if properly integrated and updated regularly and frequently.
Conceptual example of a population-level digital twin dashboard for maternal and child health, integrating care pathways, system constraints, inequity patterns, and simulated policy options to support learning and decision-making.
What do digital twins cost?
Building and maintaining a digital twin involves investments in data integration, analytics and modelling. This is not inexpensive in absolute terms, but the cost may be low relative to alternative decision-making approaches. Allowing the virtual testing of strategies might avoid costly missteps and allocate scarce resources more effectively than traditional trial-and-error or retrospective planning.
What improvements are possible?
Digital twins are not a silver bullet, but they can improve how health systems learn, prioritise, and act, resulting in:
- More precise targeting of interventions. Instead of one-size-fits-all programmes, test which combinations of actions work best in specific contexts.
- Earlier identification of emerging system risk. By integrating clinical, social, and environmental data, identify patterns of rising risk such as delayed access, service bottlenecks, or geographic clustering of adverse outcomes, allowing earlier and more effective intervention at a population level.
- Improved health-system design and resource allocation. Simulate difficult policy choices such as whether to invest in community health workers, clinic expansion or improved transport. Identify which options are most likely to reduce preterm birth, maternal morbidity or neonatal mortality in a given district.
- Increased visibility and actionability of health inequities. By modelling how social determinants interact with care delivery, move past identifying “high-risk patients” to understanding where and how the system itself creates risk; enable upstream redesign instead of downstream remediation.
- Support for continuous learning. Instead of relying on slow, retrospective evaluations, health systems can monitor changes over time, adjust strategies, and refine interventions as populations and conditions evolve.
How do we know a digital twin is “right”?
“Digital twins” in population health are not proven to be reliably predictive for policy decisions in the way a lab test is proven. Many are still research programmes, prototypes or decision-support tools with partial validation.
The minimum standard is that a digital twin can reproduce known patterns, and predict outcomes in new settings with the level of uncertainty clearly stated. Predictions can be compared against results of clinical trials and epidemiologic studies.
For population-level maternal and child health twins, start with back-testing and external validation, then evaluate whether decisions made with the twin outperform usual planning (or simpler baselines) in prospective deployments.
Without that step, a twin is best treated as a hypothesis generator rather than an authority.
Limits, risks and guardrails
To be clear, digital twins are not magic solutions, replacements for clinicians or shortcuts around political and social accountability. They do not fix underfunding, workforce shortages or structural racism.
Their risks include reinforcing bias, if models are trained on skewed data, excluding specific communities.
Population-level digital twins should be transparent and interpretable, involve communities in design and governance and be used to inform decisions.
A model is not an unchallengeable authority; accountability cannot be shifted from policy makers to algorithms. Appropriate use preserves human accountability and ethical judgement.
Image: OpenAI
A better future for maternal and child health
Maternal and child health failures are rarely caused by ignorance. We know what works. The problem is that complex systems resist simple fixes.
Digital twins offer a way to understand complexity, test solutions without harm and align clinical care with public health and social policy.
In rich countries, they can help expose and reduce structural disparities. In LMICs, they can accelerate learning while helping protect the most vulnerable.
If used well, digital twins provide system insights for broad improvement and reduction of health inequities. In maternal and child health, these insights could be lifesaving.
Readings and References
- Defining digital twins and their use in healthcare
Mulder et al. Dynamic digital twin: diagnosis, treatment, prediction, and prevention of disease during the life course. J Med Internet Res (2022)
https://www.jmir.org/2022/9/e35675
Digital twins are virtual replicas that combine real data with simulation models to support decision-making in complex systems.
Papachristou et al. Digital Twins’ Advancements and Applications in Healthcare, Towards Precision Medicine. https://www.mdpi.com/2075-4426/14/11/1101
Digital twins in healthcare represent models that encapsulate both clinical and physiological characteristics to support interpretation, prediction, and decision-making.
Pitt et al. Systems modelling and simulation in health service design, delivery and decision making (May 2016). https://qualitysafety.bmj.com/content/25/1/38
A range of modelling and simulation methods have been used in health services research. These approaches can support planning and policy decisions across multiple levels of healthcare delivery but require stronger integration with real-world data and decision processes to improve service design and outcomes.
- Digital twins in healthcare research
Machado & Berssaneti. Literature review of digital twin in healthcare. Heliyon. 2023 Aug 24;9(9):e19390 https://pmc.ncbi.nlm.nih.gov/articles/PMC10558347
Bibliometric (literature) review. Digital twin technology in healthcare is increasingly studied but still immature. Many applications are described but challenges in integration and validation are noted.
- Examples of digital twin specific implementations
Rudsari et al. Digital twins in healthcare: a comprehensive review and future directions. Front. Digit. Health, 18 November 2025 https://www.frontiersin.org/journals/digital-health/articles/10.3389/fdgth.2025.1633539/full
Nov 2025 review. Digital twin applications achieving measurable clinical successes (e.g., disease prediction accuracy), while also emphasising challenges like data integration and validation.
- Maternal & fetal health digital twins
Calcaterra et al. Maternal and fetal health in the digital twin era. Front. Pediatr., 12 October 2023 Sec. Neonatology Volume 11 – 2023
https://www.frontiersin.org/journals/pediatrics/articles/10.3389/fped.2023.1251427
Peer-reviewed evidence describing digital twin systems in maternal–fetal health that can continuously update records and support early detection of pregnancy-related risks.
- Global maternal mortality statistics
World Health Organization, “Maternal mortality” fact sheet.
https://www.who.int/news-room/fact-sheets/detail/maternal-mortality
WHO’s 2025 maternal mortality fact sheet. Global distribution of maternal deaths showing that almost all occur in low- and lower-middle-income countries.
WHO infant & neonatal mortality:
https://www.who.int/data/gho/data/themes/topics/infant-and-young-child-mortality
- Digital Twins for Health Society