HYDROCEPHALUS

BACKGROUND/PROBLEM STATEMENT

Hydrocephalus (HYC) is the segregation of fluid in the cavities (ventricles) deep within the brain. The excess fluid increases the size of the ventricles and puts pressure on the brain. Cerebrospinal fluid normally flows through the ventricles and bathes the brain and spinal column. Hydrocephalus is a frequent complication which is affecting almost all age groups following subarachnoid haemorrhage. Few studies investigated the association between laboratory parameters and shunt-dependent hydrocephalus. Non-contrast materialenhanced head computed tomographic (CT) examination is an important method for the diagnosis of HYC because it can observe the enlargement of the ventricles, and sometimes determine the cause of HYC. However, due to the lack of uniform standards, different range of patients’ ages and the various levels doctors’ expertise, it is rather difficult to reach a diagnosis. Therefore, after a detailed analysis and research, Ai labs developed and implemented a Proof of Concept (POC) using its proprietary engine Minsky™ to highlight the need for a shunt for an individual and mortality caused by Hydrocephalus.

KEY BENEFITS (MINSKY™)

• User-Friendly cloud based AI platform
• Scalable across various domains/data.
• Provides you a list of % dependency features that can be used to optimize the outcomes.
• Ability to fine tune or optimize the models by trying different algorithms / prediction attributes
• Easy integration with other third party solutions such as TABLEAU for data visualization

RESULTS/BENEFITS

• Can effectively identify the characteristics of Hydrocephalus by reading CT images of the brain and analyzing the factors of the shape and volume of ventricle etc. for diagnosing hydrocephalus.
• Ability to provide auxiliary diagnosis of image results and reduce the burden on untrained Physicians.
• Improved patient health
• Can predict early the requirement of a shunt for an individual looking at current diagnosis so as to avoid catastrophic health issues.<./li>

SOLUTION OVERVIEW:

After comprehensively evaluating the client‘s challenges/data, we used Minsky™ to accurately model the historical data which comprised patient data across an age group of 30 to 65 years. The patient data also included parameters such as Age, Gender, CT Scan images, Last Volume, WBC, Lymphocytes, Monocytes, Albumin, Sodium, Glucose, Potassium Level, Platelets etc.
Once the detailed modelling was performed by Minsky™ for the selected Algorithms, historical data used by Minsky™ was separately analysed to predict a) the requirement for a Shunt and b) Shunt Mortality for those patients that required a shunt.

Project Results:

Minsky™ Results for Shunt Requirement:

Based on the data from Fig 1 above, our analysis shows that Stochastic Gradient Descent Classifier (82.2%), Gaussian Naïve Bayes (81.8%) and Logistic Regression (80.3%) gave the highest top three model accuracies. In terms of the patient data, the Last Volume showed the highest dependency for the shunt requirement followed by Sodium and Osmolality levels.

Minsky™ Results for Shunt Mortality:

Based on the data from Fig 2 above, our analysis shows that Random Forest Classifier (91.2%), Gradient Boosting Classifier (88.2%) and Logistic Regression (88.2%) gave the highest top three model accuracies. In terms of the patient data, Creatinine showed the highest dependency to predict Shunt Mortality followed by Lymphocytes and Bun levels.