Personalised Modelling for Stroke Risk Prediction

Abstract

The Brain-Like Artificial Intelligence (BLAI) is pioneered by Prof.Nikola Kasabov and here it is applied to a specific application.

This project develops a novel personalised modelling method and system aimed at developing computational prognostic or diagnostic systems for risk of stroke. The proposed method is based on integration of different data processing techniques for appropriate selection of features (variables) and neighbouring samples. This has the potential to identify the most important characteristics of an individual through personalised profiling and improves the classification/prediction of output (risk of stroke event) as compared to global modelling. In this section, instead of building a global model and training it with the whole population in data, for every person we will build a personalised SNN model (PSNN) to train it only on a subset of data which belongs to individuals who have similar integrated static clinical factors and dynamic environmental data.

The proposed NeuCube personalised modelling is performed based on the following steps:

  1. For a new person data xi
    1. A cluster of K number of patients (had stroke) who have small distance to xi will be selected (according to static data of stroke). Clustering of integrated static-dynamic data is performed using the new algorithm dWWKNN (dynamic weighted–weighted distance K-nearest neighbours). For a new individual xi, we rank the contribution of each of the k neighbouring samples based on integrated static-dynamic distance to the xi, giving greater rank to closer neighbours.
    2. For every member in this cluster, the date of stork will be found and two environmental samples will be extracted with respect to the below time intervals:
      1. High risk of stroke: environmental data of 20 days (a parameter that can be adjusted by the user) before the stroke event
      2. Low risk of stroke: environmental data of 20 days in the interval of 60 to 40 days before the stroke event
  2. K*2 environmental samples will be used for creating a personalised SNN model, training and testing.
  3. The output prediction provides a probability whether the person xi is in high risk of stroke or low risk.

FIGURE 1. Block diagram of the NeuCube personalised modelling for prediction of risk of Stroke using both static and dynamic environmental data.

FIGURE 2. Representation in a NeuCube model trained on environmental data related to a sub-group of people who suffered a stroke (20 days before the stroke event happens).

Related Papers, Patents and Benchmarking

The proposed methods and systems, when compared with traditional statistical and machine learning methods, showed superior results in the following aspects:

  1. Identification of the most important characteristics of an individual through personalised profiling;
  2. Improved classification/prediction of output (risk of stroke event) as compared to global modelling;
  3. Better visualisation of the created models, with a possible use of VR;
  4. Better understanding of the data and the processes that are measured;
  5. Enabling new information and knowledge discovery through meaningful interpretation of the models.

See also some of the related papers:

Kasabov, N. K. (2014). NeuCube: A spiking neural network architecture for mapping, learning and understanding of spatio-temporal brain dataNeural Networks52, 62-76.

Kasabov, N., Feigin, V., Hou, Z. G., Chen, Y., Liang, L., Krishnamurthi, R., Othman, M., Parmar, P. (2014). Evolving spiking neural networks for personalised modelling, classification and prediction of spatio-temporal patterns with a case study on stroke. Neurocomputing, 134, 269-279. doi:10.1016/j.neucom.2013.09.049

Kasabov, N., Feigin, V., Hou, Z. -G., Chen, Y., Liang, L., Krishnamurthi, R., Parmar, P. (2014). Evolving spiking neural networks for personalised modelling, classification and prediction of spatio-temporal patterns with a case study on stroke. Neurocomputing, 134, 269-279.

Feigin, V., P.Parmar, S. Barker-Collo, D.A Bennett, C.S Anderson, A. G Thrift, B. Stegmayr, P. M Rothwell, M.Giroud, Y. Bejot, P. Carvil, R.Krishnamurthi, N.Kasabov (2014). Geomagnetic Storms Can Trigger Stroke. Stroke, STROKEAHA-113.

N.Kasabov, V.Feigin, Z.Hou, Y.Chen, Improved method and system for predicting outcomes based on spatio/spectro-temporal data, PCT patent WO2015/030606 A2, US2016/0210552 A1. Granted/Publication date: 21 July 2016.

N.Kasabov, Data Analysis and Predictive Systems and Related Methodologies, US patent 9,002,682 B2, 7 April 201


R&D System

For this project, an R&D system has been developed based on NeuCube. The system can be obtained subject to licensing agreement.


Developer

The developer of this project is:

Maryam Gholami Doborjeh