DFKI

Resources · Pillar Guide

Medical Device Development:
From Research to AI-Enabled Prototype

Developing a new medical device increasingly requires more than hardware or software alone.

Modern medical technologies combine sensing, physiological data, signal processing, medical imaging, machine learning and clinical requirements into a single system.

DFKI helps healthcare and MedTech organisations investigate whether ambitious technology concepts are technically feasible — and turn promising ideas into validated prototypes and research demonstrators.

Discuss Your Medical Device Idea Explore Our Research Expertise
DFKI researcher and clinician reviewing a research workstation in a clinical environment

From Idea to Working Technology

The Difficult Part Is Connecting the Pieces

A medical technology concept may begin with a clinical problem, a new sensor, an imaging modality, an existing dataset or an idea for using AI.

The challenge is turning those individual components into a system that actually works.

This is where applied research becomes particularly valuable.

Before committing to full-scale product development, organisations can investigate critical technical questions, test alternative approaches and validate the core technology through focused R&D.

  1. 01Clinical Need
  2. 02Sensors / Imaging / Data
  3. 03Signal & Image Processing
  4. 04AI / Machine Learning
  5. 05Prototype
  6. 06Technical Validation
  7. 07Clinical Development

R&D Path

A Research-Driven Medical Device Development Process

01

Define the Clinical and Technical Problem

Clarify what the device should measure, detect, predict or support and identify the technical uncertainties that must be solved.

02

Feasibility Study

Evaluate available sensors, data, algorithms, imaging technologies and system architecture.

Key question: Can the idea work technically?

03

Data, Sensors & Signal Processing

Develop the data acquisition and processing pipeline.

  • Physiological signals
  • Wearable sensors
  • Imaging data
  • Multimodal data
  • Biosignals
  • Sensor fusion

04

AI & Algorithm Development

Investigate suitable machine learning, deep learning or analytical approaches.

  • Classification
  • Detection
  • Segmentation
  • Prediction
  • Anomaly detection
  • Temporal modelling
  • Multimodal AI

05

Prototype / Proof of Concept

Integrate the relevant hardware, data processing and AI components into a working research prototype.

06

Technical Validation

Evaluate performance, robustness and limitations before proceeding toward larger clinical or commercial development.

The exact development path depends on the technology, clinical application, available data and maturity of the concept.

You Don't Need to Start With a Finished Product Specification

R&D projects often begin much earlier.

A Clinical Problem

“We believe technology could solve this problem, but we don't yet know how.”

A Sensor or Hardware Concept

“We can measure something interesting, but need to understand how to turn the signal into clinically useful information.”

Existing Data

“We have valuable clinical or physiological data and want to investigate what AI could extract from it.”

An AI Concept

“We have an AI use case but need to determine the right data, architecture and validation approach.”

An Existing Medical Device

“We want to add sensing, intelligence or AI capabilities to an existing product.”

A Research Question

“We need to establish technical feasibility before committing to product development.”

Healthcare AI R&D

Research Expertise Across the Medical AI Technology Stack

DFKI's healthcare AI research spans the complete chain from physical sensing and medical data to advanced AI methods — from sensing to AI.

This allows research questions to be approached as complete systems rather than isolated algorithms.

Layer 2 · Data & Signals

Layer 3 · Artificial Intelligence

Layer 4 · Applications

Questions We Can Investigate Together

  1. Q01Can a wearable sensor reliably capture the physiological signal we need?
  2. Q02Can AI identify clinically meaningful patterns in the signal?
  3. Q03Can multiple sensors or data modalities improve prediction accuracy?
  4. Q04Can medical images be analysed automatically?
  5. Q05Can longitudinal imaging quantify disease progression or treatment response?
  6. Q06Can an existing medical device be enhanced with AI?
  7. Q07Can we detect deterioration before conventional clinical indicators?
  8. Q08Can a research algorithm operate reliably with real-world patient data?
  9. Q09Can a technical concept be demonstrated before committing to full product development?

Medical Device Development Across Multiple Technology Areas

Medical Imaging AI

Image analysis, segmentation, registration and temporal analysis.

Explore expertise

Wearable & Sensor Systems

Continuous physiological monitoring and intelligent sensing.

Explore expertise

Medical Signal Processing

Extract clinically relevant information from complex physiological signals.

Explore expertise

Remote Patient Monitoring

Combine connected devices, data processing and AI for continuous monitoring.

Explore expertise

Multimodal Healthcare AI

Combine imaging, sensor, clinical and other data sources.

Explore expertise

Clinical AI

Develop and investigate AI methods for real-world healthcare applications.

Explore expertise

Reduce Technical Uncertainty Before Scaling Development

One of the most expensive mistakes in medical technology development is committing too early to a specific technical architecture.

Applied R&D can answer fundamental questions before significant resources are committed to engineering, clinical development and regulatory activities.

Test Feasibility

Determine whether the underlying technical concept works.

Compare Approaches

Evaluate alternative sensors, algorithms or system architectures.

Build Evidence

Generate technical results that support the next development decision.

Applied Research

From Research Questions to Working Technology

DFKI × Xenoma

AI Wearable Technology

Challenge
Analysing human movement continuously and outside the lab.
Research approach
Intelligent wearable sensor technology with real-time movement analysis.
Technology
Smart textiles, wearable sensors, machine learning
Research result
Research on real-time human motion analysis relevant to rehabilitation and digital health.
Learn more

DFKI × Hitachi

AI for Occupational Health

Challenge
Quantifying physical workload and ergonomic risk.
Research approach
AI-powered wearable sensing to measure workload and human movement.
Technology
Wearable sensing, activity recognition, AI
Research result
Applied research towards improved occupational health and safety.
Learn more

DFKI × Hitachi × Xenoma

Intelligent Sensor Suit

Challenge
Recognising human activity in real time from the body.
Research approach
Combining AI, wearable sensing and smart textiles in one system.
Technology
Sensor suit, sensor fusion, human activity recognition
Research result
A research collaboration demonstrating real-time movement analysis.
Learn more
See all case examples

When DFKI Can Be the Right R&D Partner

This page is particularly relevant for:

  • MedTech companies developing new products
  • Medical device manufacturers adding AI capabilities
  • Healthcare technology companies working with sensors or physiological data
  • Imaging technology companies
  • Digital health companies moving toward physical medical technology
  • Pharmaceutical and life-science organisations developing monitoring technologies
  • Corporate R&D and innovation teams investigating new healthcare technologies

Typical project stages

  1. Idea
  2. Feasibility
  3. Research
  4. Proof of Concept
  5. Prototype
  6. Validation

DFKI is especially valuable in the early stages, while important technical questions are still unresolved.

Discuss Your R&D Challenge

Have a Medical Device Idea or Technical Challenge?

You don't need a finished specification.

If you are exploring a new medical device, sensing technology, healthcare AI application or intelligent monitoring system, we can start by understanding the technical question you are trying to solve.

Discuss Your Idea Join the Healthcare AI Community

A first conversation can focus simply on the technology, available data and the key feasibility questions.