It was a pleasure witnessing the participation of Kinseed in research conducted at UCL during the CHIMERA webinar “Leveraging AI To Support Clinical Decision-making during Emergency Transport of Critically Ill Children to PICU.” The webinar was presented by Dr Padmanabhan Ramnarayan, Consultant in Paediatric Intensive Care and Retrieval, who was joined by Bob Hundal, our Chief Technology Officer (CTO), in the panel discussion.

Below are some key take away points from the webinar…

Kinseed’s SwiftCare solution is used extensively across various UK paediatric critical care transport services, including the CATS team at Great Ormond Street Hospital. When a critically ill child is transported to hospital in an ambulance, they’re connected to one of our devices. These devices enable doctors to monitor the patient’s real-time condition and provide necessary care, regardless of their location.

Our MediConnect application is connected to the monitor, which then sends output to the MediVue application where real-time vitals can be viewed. Clinical teams also use MediLog, a cloud-based server, to document comprehensive patient information. The MediLog application is accessible on mobile devices, offering an offline feature to prevent data loss in areas with poor connectivity during transport.

In addition to reporting patient vitals, Kinseed’s SwiftCare solution incorporates an AI model that aids risk analysis within healthcare settings. This AI model serves as a valuable tool for supporting clinical decision-making processes.

 

Here’s a breakdown of each step:

1

Data pre-processing (pattern recognition):

This step involves preparing the input data for analysis by applying various pre-processing techniques. Pattern recognition algorithms may be used to identify and extract relevant patterns or features from the raw data

2

Extract physiotypes (multivariate time series analysis):

Once the data is pre-processed, multivariate time series analysis techniques are applied to extract physiotypes. Physiotypes refer to specific physiological patterns or characteristics obtained from the collected data. These physiotypes could include vital signs, biomarkers, or other relevant physiological parameters.

3

Real-time explainable risk analysis (Risk assessment using edge AI):

Using the extracted physiotypes, real-time risk analysis is performed. This involves assessing the risk level or probability of certain outcomes or events based on the patient’s physiological data. The AI model analyzes the data and provides insights into the patient’s condition, such as predicting the likelihood of developing complications or identifying high-risk situations.

4

Tackle bias and practicality (Risk assessment using edge AI):

In this step, efforts are made to address bias and ensure the practicality of the AI model. Bias can arise due to imbalanced datasets or inherent biases in the data collection process. By carefully considering and addressing these biases, the model aims to provide fair and accurate risk assessments. Additionally, the use of edge AI refers to deploying the AI model directly on edge devices or within the clinical setting, enabling real-time analysis and decision-making without relying solely on centralised systems.

5

Implement on edge and centre (clinical expert system):

Finally, the AI model is implemented both at the edge (on devices or within local infrastructure) and at the centre (cloud-based system). This implementation allows for the deployment of the AI model in clinical environments, where it can serve as a clinical expert system. The system provides insights and recommendations based on real-time analysis of patient data, assisting healthcare professionals in making informed decisions and improving patient care.

Overall, this model encompasses various stages of AI model development, including data pre-processing, feature extraction, real-time risk analysis, bias handling, and the implementation of the model in both edge and central systems to support clinical decision-making and improve patient outcomes. The model displays how transport teams can leverage AI to support clinical decision-making during emergency transport of critically ill children to PICU. The use of AI plays a pivotal role in significantly improving clinical decision-making and the overall effectiveness of paediatric critical care transport.

As a company, we take immense pride and joy in being actively involved in groundbreaking research at UCL. It was a true pleasure for us to showcase our commitment and expertise during the webinar. Witnessing the impact of our contributions firsthand was both gratifying and inspiring, further fueling our dedication to advancing innovation in collaboration with esteemed institutions like UCL.