OsteoPorosis Treatment Identification Using Machine Learning — OPTIMAL  

Osteoporosis Clinical Trial

Official title: OPTIMAL - OsteoPorosis Treatment Identification Using Machine Learning

Status Not yet recruiting

Clinical Trial Summary

OPTIMAL is a pilot feasibility study for a machine learning (ML) based enhanced screening software for osteoporosis.

This tool has been created using machine learning, based on data from NHS Greater Glasgow and Clyde. The study will contact individuals deemed at high risk by the study (750 patients will be re-identified, and these will be contacted starting from the highest risk until 250 patients are recruited) and perform DXA scans, clinical review, and bloods tests that are relevant to osteoporosis. This data will then be compared to the predictions made by the OPTIMAL enhanced screening tool, in order to test how effective it is.

Clinical Trial Description

The OPTIMAL trial is a feasibility pilot study for an enhanced screening method using a tool developed using machine learning. This tool has been created by Lenus Health Ltd using anonymised health data from NHS GG&C of patients between the age of 50 and 80 as of 1st of January 2010.

The validation cohort will include all patients in GG&C between the ages of 50 and 80 as of 1st January 2022 who are not known to have a diagnosis of osteoporosis.

Nanox (Israeli AI company) will use their existing platform to analyse de-identified CT scans of the 5000 patients identified as highest risk in this population to asses for vertebral fracture, which will be used as a parameter in risk assessment for osteoporosis. Because of technical limitations, only a selection of patients can have their imaging assessed by the Nanox platform. The 5000 highest risk patients will be reidentified so that their images can be extracted and de-identified for analysis. We expect approximately 30% of these patients will have available imaging. Identifiable data will remain within the GG&C cloud.

The trial will use the software to screen for patients at high risk within a cohort of patients aged 50-80 without a diagnosis of osteoporosis. DXA scanning will be used as a ground truth. Patients will have DXA scanning as part of the trial.

There will be a clinical review to remove patients that are felt to be inappropriate for the trial, such as people approaching the end of life, or unable to give consent. The reason for exclusion will be recorded to allow auditing and assessing for bias in selection.

Patients, starting with those at highest risk, will then be contacted by the Fracture Liaison Service until 250 have been recruited. This number has been reached based on power calculations to ensure this trial can effectively compare the OPTIMAL tool with FRAX. This contact will be in the form of a letter from the fracture liaison team, and will come with a patient information sheet and consent form. These forms have been created with the input of a patient experience group.

The participant, should they agree to partake, will then be invited for a single clinic visit at the Queen Elizabeth University Hospital.

At this visit, a doctor will review the patient, and ask questions about their health. They will have blood tests which are relevant to osteoporosis, as well as general health. They will also have a DXA scan to confirm if they do or do not have osteoporosis. A FRAX score will be generated for the purposes of comparison.

Upon completion of data cleaning, at the end of the trial, all data will be transferred The University of Strathclyde secure servers for analysis by Dr Conor McKeag, with oversight from Dr David Young to assess the efficacy and accuracy of the OPTIMAL tool. This will be compared to FRAX, the current standard for risk prediction in osteoporosis. ;

Related Conditions & MeSH terms

• Osteoporosis
• Osteoporosis Risk

• NCT number: NCT05678569
• Study type: Observational
• Source: NHS Greater Glasgow and Clyde
• Contact: Ruairidh Davison
• Phone: 01414516869
• Email: Ruairidh.Davison@ggc.scot.nhs.uk
• Status: Not yet recruiting
• Start date: February 2023
• Completion date: June 2023