Medical technology (MedTech) manufacturing spans a wide spectrum of products, from simple everyday tools to highly complex, lifesaving machines. It also operates under extreme regulatory oversight because of its impact on human lives. Digital Device History Records (eDHRs) are the norm across MedTech manufacturing today, but many organizations still struggle with inconsistent ways of modeling and exchanging unit level data across operational systems such as MES, QMS, ERP and PLM systems.
This lack of alignment across systems can limit automation, slow investigations and create data integrity challenges. As the MedTech industry adopts more intelligent forms of automation-including early elements of physical artificial intelligence (AI)-the need for greater consistency in how device level information is structured is becoming more important. This article explores the consistency gap and why it exists, how it affects modern manufacturing and a practical framework that could support more unified unit level traceability (ULT).
eDHRs exist, but fragmentation remains
Over the past decade, manufacturers have shifted from paper DHRs to digital systems capturing processes, materials, approvals and quality events. While this has reduced administrative burden and improved accessibility, it has not created uniformity.
Different vendors and sites model unit level data in various ways-ranging from structured databases to PDFs, uploads or custom tables. Unit histories often need to be manually reconstructed and exception handling may not follow the specific device configuration. Approvals may reference documents or lots rather than the specific unit or lot they actually apply to. This creates several operational challenges:
- Time consuming investigations
- Approval drift during engineering changes
- Exceptions not consistently tied to a specific unit
- Analytics that require extensive data cleansing
- High effort system integrations that are difficult to validate
From an operations technology standpoint, these issues appear frequently during deviations, change control and audits. Digitization improved accessibility, but not semantic consistency.
A key reason fragmentation persists is that most legacy systems were designed independently, long before modern interoperability expectations. As a result, each system evolved its own data model, terminology and event semantics-making alignment difficult even after digitization.
Existing standards don’t fully address unit level semantics
MedTech already relies on strong standards, including:
- ISA 88 for batch and procedural control
- ISA 95 for system integration
- 21 CFR Part 820 for required DHR content
- ISO 13485 for quality management
- 21 CFR Part 11, which governs electronic records and signatures, is also highly relevant. Part 11 requires systems to log who performed each action, when it occurred and how records are protected-capabilities that directly support unit level traceability when combined with a structured ULT framework.
However, these standards do not describe how to digitally structure unit level data across systems. They define process, integration layers and documentation expectations, but not how individual device records should be modeled in real time.
Manufacturers, therefore, create their own representations for device identity, event meaning, genealogy, state changes, material consumption, deviations and holds and signatures and approvals.
Because each organization does this differently, interoperability becomes challenging-particularly as automation grows more sophisticated.
ISA 112, focused on SCADA systems, is also increasingly relevant because SCADA often acts as the real time data collection layer feeding DHR systems. Including ISA 112 concepts can help ensure consistent event capture from the factory floor.
Costly ambiguity
Although the industry is not yet deploying fully autonomous systems, MedTech is steadily adopting technologies that blend sensing, logic and real time decision making. This includes:
- Equipment with embedded intelligence
- Digital work instructions tied to operator qualifications
- Vision based inspections that adjust to variation
- Analytics that correlate patterns across product and process
- Digital twins for simulating changes before implementation
These represent early elements of Physical AI-automation that perceives, decides and acts in the physical environment. As the systems become more capable, they require more dependable and consistent unit level data.
For example, a robot cannot interpret an unstructured genealogy document. An AI model cannot infer meaning from inconsistent event definitions. A digital twin cannot synchronize against device data that varies by system. The smarter automation becomes, the more costly the ambiguity becomes. This is driving renewed interest in more unified unit level traceability.
Start with a Unit Level Traceability framework
A possible starting point is a Unit Level Traceability (ULT) framework. This is not intended as a formal standard but rather as a conceptual structure that could help guide the industry toward greater consistency.
Traceability granularity is itself a risk-based decision: implantable or life-sustaining devices typically require unit-level control under 21 CFR 820.65, while other devices are appropriately traced at the lot or batch level. The framework below flexes to whichever granularity a manufacturer's risk classification and DMR require.
It starts with a canonical set of data objects. This shared set of core objects could help systems “speak the same language” and understand each other more reliably. Examples include:
- DeviceUnit – control-number identity (unit, lot or batch, per the DMR), configuration, version
- OperationStep – actions performed
- ProcessParameters – measured values and limits
- MaterialLot/Genealogy – components and substitutions
- Equipment & Tooling – machines and calibration context
- Personnel & Qualifications – operators and authorizations
- Exceptions – deviations, holds, scrap, rework
- eSignatures & AuditEvents – approvals tied to specific states
Each event affecting a device could benefit from capturing atomic event semantics. These include: what happened; who performed or authorized it; when and where; materials or equipment used; and the before/after device state. This supports clearer investigations, analytics and audits.
For example, a MES-to-QMS interoperability profile could define how a deviation raised during assembly is automatically linked to the exact unit or lot involved, its configuration, the operator and the equipment used-ensuring consistent meaning across both systems.
Genealogy captured and updated in real time-not assembled later-enables faster containment and clearer visibility into rework and component usage. Exceptions and approvals would be linked to unit state. Binding exceptions and signatures to the exact configuration at the moment they occur may help reduce ambiguity and prevent approval drift during engineering changes.
How manufacturers can begin moving toward ULT
Even without a formal standard, organizations can start with steps such as:
- Mapping current eDHR elements to ULT like objects
- Converting genealogy from documents to structured data
- Linking deviations and approvals to specific unit configuration
- Moving from batch exports to real time updates
- Piloting a single interoperability profile (such as MES to QMS for deviations) These actions can improve data quality today while preparing for increasingly intelligent automation tomorrow.
Conclusion
As MedTech manufacturing evolves toward more advanced automation-including early forms of Physical AI-the limitations of fragmented unit level traceability are becoming more visible. While eDHR digitization improved accessibility, it did not create consistent semantics or models across systems.
A more unified approach to unit level traceability could support higher data integrity, simpler integration, stronger audit readiness, and safer automation. The ULT Framework outlined here is not a prescriptive standard, but a possible foundation for ongoing conversation within the industry as it prepares for the next decade of digital and physical transformation.
ISA training that supports digital device history records
- ISA 95 Enterprise Control System Integration: Understanding the network infrastructure for connectivity from the plant floor to the enterprise, which, when done correctly, provides a rudimentary level of cybersecurity.
- ISA 88 Batch / Procedural Control (IC40): Knowledge for building or configuring MES workflows, designing eDHR execution logic, structuring eBR/eDHR data capture, and implementing review by exception logic control.
- ISA/IEC 62443 Cybersecurity Fundamentals (IC32): With any IIoT and information that flows to and from the Internet, an opening for bad actors is created. The IC32 Cybersecurity track helps designers of the proposed system put together a secure implementation.
- Manufacturing Operations Management (MOM/MES Oriented Courses): As DHR systems rely on understanding structured MOM and MES systems, this training supports core behaviors such as dispatching work orders, enforcing step execution, capturing electronic signatures, handling deviations/exceptions, and managing genealogy and traceability in real time.
