Our Services › ClinAssIst - AI assistant for clinical data
ClinAssIst - AI assistant for clinical data
Signifikans is developing an AI assistant to process clinical data (launch 2027).
Signifikans is developing an AI assistant to process clinical data.
Examples of clinical problems to solve (and more):
Drug Safety: Patient safety evaluation from clinical trials - support to data monitoring committees. Input data will be patient listings on laboratory data, adverse events and concomitant medication
Clinical Trial Data Management: Assigning participants to statistical analysis populations and identifying protocol deviation - support to blind data review. Input data will be patient listings and criteria of protocol deviations
Clinical Trial Data Management: Establish pattern of data queries from site on ongoing clinical trials to improve quality of database setup - support to data management. Input data will be queries generated from site in the EDC system.
Quality assurance: Signal detection for control of risk to maintain GCP (R3) - support for your clinical trial team. Input data TBD.
The AI assistant will run on a closed network on our own standalone server - this will ensure highest level of security and performance (processor: AMD Ryzen 5 7400 CPU, 64 GB DDR5 ram. Graphics card: Nvidia RTX 5090 m. 32 GB ram).
The LLM model we use for now is Qwen 3.8-27B open-weight model that runs in several loops. Some loops for output generation and some loops for cross-check of output vs specification.
The models will be given a set of instructions (skills or SOPs) to guide the process.
After each process loop, experience and lessons learned will be stored and used for next loop - thereby the model run will constantly improve.
Model parameters are fixed and no data will be used to change them. But the instructions to the model will constantly be improved to optimize the process.
After end task, all input data will be deleted, so no data storage on ClinAssIst.
Our GCP and GDPR compliant files storage repository CLIN-LINK will be the data input and output storage repository.
The AI assistant will be compliant to GCP, GDPR and the EU AI act.
For model validation, we will validate the output against output from traditional data analysis and compare results using Jaccard similarity index.
You will benefit from ClinAssIst:
if you have a clinical question
If you have clinical data
If data security is important to you
If cost is important to you
At launch, there will be capacity for one client at the time.
Estimated launch and booking start will be beginning of 2027.
For more information please contact Signifikans.