CVA Prediction and Prevention A.I.

CV-AI

CV-AI is made for patients who fit into the risk category group for having a stroke. CV-AI is not made to be a conclusive tool but rather an indicative model of which year the patient is at its highest risk of having a stroke.

Through the use of Neural Networks and A.I. CV-AI takes into account any and all variables of a patients personal life that increases the risk of having a stroke. This includes but is not limited to lifestyle, profession, environmental and family. Based on this information and the data of the scan CV-AI performs a 10 year based model on the patients brain to map out and predict the probability range of having a stroke in those years. From there our system will also provide recommendations for treatment and medication to reduce that probability. Different models are created for the patient to look at and for the doctor to decide what is best for the patient with the data provided.

Predicative Modeling

Preventive Modeling for Risk Category Patients

CV-AI is made for patients who fit into the risk category group for having a CVA. CV-AI is not made to be a conclusive tool but rather an indicative model of which year the patient is at it’s highest risk of having a stroke.

Once the patient has conducted their scan at any hospital, the data will be sent over to any DX lab that will then conduct analysis. CV-AI provides up to a 10 year based predicative model with additional included variables such as lifestyle, stress factors and potential environmental changes. From this we provide a range of probability with the brain-scan modelling in order to illustrate the probability of having a stroke in those 10 years.

Prevention Oriented

Preventing irreversible damage

A stroke causes irreversible damage and is one of the most debilitating things to happen to a patient. CV-AI aims at preventing patients from having a stroke early on by providing creating key models that allow for early preventive treatment and lifestyle changes.

CV-AI analytics take between 24-48 hours to process after a scan has been conducted, allowing the patient to know which course of action they have to take the same week they see their doctor. It is not dependant on any specific hospital as all data and billing gets processed directly through insurance providers and sent to a DX Lab.

Suggested Treatment Protocols

A.I. Assisted treatment protocol suggestions

DiagnostiX has an epidemiological database of more than 1 Billion total entries. Our database focuses on diagnosis for each type of patient.

The diagnosis on which DiagnostiX is trained will therefore include pre-made templates on a per patient basis for possible treatment protocols that suit best for each individual patient. This is done by cross referencing the potential individual variables with the probability of success based on previous data.

CV-AI is able to detect and create preliminary models with degrading accuracy up to 10 years.

Ischemic Stroke
Ischemic Stroke
Hemorrhagic Stroke
Hemorrhagic Stroke
Embolic
Embolic stroke
TIA
TIA (Mini Stroke)
model1
Y2Y
Ischemic Stroke

Ischemic Stroke

Sensitivity (% 65.7 vs. %31.4, P value = 0.001), positive predictive value (% 85.2 vs. % 65, P value = 0.03)
Hemorrhagic Stroke

Hemorrhagic Stroke

Sensitivity (% 68.7 vs. % 31.4, P value = 0.001), positive predictive value (% 86.5 vs. % 65, P value = 0.03)
Embolic

Embolic Stroke

Sensitivity (% 65.7 vs. %31.4, P value = 0.001), positive predictive value (% 85.2 vs. % 65, P value = 0.03)
TIA

TIA (Mini Stroke)

Sensitivity (% 68.7 vs. % 31.4, P value = 0.001), positive predictive value (% 86.5 vs. % 65, P value = 0.03)
Stroke Model

Y2Y

Year based modeling on a 10 year predicted scale positive predictive value max 65%

CV-AI quantification and validation is still currently undergoing but early internal studies have shown a current accuracy rate of 85% at a max probability range of 65%, with more than 1 million datasets being incorporated for training.

All 12 models had excellent discrimination and calibration, with 8 of 12 meeting prespecified performance criteria (AUC ≥0.8, mean absolute error ≤0.4).

More Info

Drop us a line at support@diagnostix.ai to know how we can set up DiagnostiX in your imaging workflow.