The idea of using the powerful super-computer, will enable the Danish scientists to have a rough AI engine model available in about three weeks.
There is a race against time around the globe to manufacture ventilators in large quantities so as to face the onslaught of COVID-19 infection, with India being no exception in this effort. Defence PSUs, Ordnance Factory Board (OFB) and Defence Research and Development Organisation (DRDO)and academia have joined hands with the private sector to ensure that the ventilators are available and are in working conditions in the hospitals across the country.
What are Ventilators?
Ventilators helps a patient in getting more oxygen into their lungs and take the carbon dioxide out. The demand for more ventilators in today’s COVID-19 pandemic is high not because they are a cure but to assist lung function while the patient is critically infected and recovering.
What are WHO Specifications?
There should be a flexible breathing circuit, a control system, monitors, and alarms.
There should be specialized breathing circuits, oxygen accumulators.
Should have humidifiers or heat and moisture exchangers (HMEs).
The devices use positive pressure to deliver gas to the lungs at normal breathing rates.
There is also an endotracheal tube, a tracheostomy cannula, or a mask.
Power supply is through a line from an internal or external battery.
Predictive Artificial Intelligence Model
In the midst of the global pandemic University of Copenhagen have innovatively used Artificial Intelligence (AI) technique to create models to calculate and forecast the requirement of intensive care and ventilator support for patients in early stages of COVID-19 infection. The effort at Copenhagen is to predict cases requiring ventilator within a week etc. This makes it feasible to optimise the use of ventilators, especially when their availability is limited and with more patients to be tended to.
Milind Kulshreshtha, C4I expert says, “Various efforts to optimise the ventilator machines are being explored including trials of mechanical distributors to link up more ventilators to a source etc.”
In the Indian context
“Every AI system is as good as the data we put into developing the solution and, in the Indian context, this data is yet to be discovered. Due to issues like data-set biases, AI tool developed for a specific population may not be found robust when applied to another group. Maybe an ab-initio development of an AI system with local data could be an exercise which may serve more than an academic interest in weeks to come,” Kulshreshtha explains.
According to him the initial effort by medical researchers at Denmark is to identify patterns from Danish COVID-19 patients who have crossed multiple stages of infection till now. And these Data pertaining to the patients is being sourced from multiple agencies for the Algorithms to be ‘trained’ using AI techniques. “The purpose is to confirm if the computer can recognize unique patterns which humans are unable to figure out on their own. These patterns shall be compared with electronic health records of the newer patients hospitalized so that the super-computer may predict the likelihood of ventilator requirement for a particular patient. Endeavour here is to ‘train’ and ‘test’ the model to calculate advance notice available before the patient is required to be shifted on to a ventilator,” the C4I expert explains.
The idea of using the powerful super-computer, will enable the Danish scientists to have a rough AI engine model available in about three weeks. However, they do also explicitly warn that this AI tool is primarily focussed for ventilator planning and is not at all a basis for treatment. The success of this effort shall provide a large benefit to Medical staff to improve their service to the patients, especially when ventilators are limited.