COVID Algorithm - Futurshealth

Cases of Good Practice

Using AI, our algorithm can
detect the clinical deterioration of
patients diagnosed with Covid 19.

The challenge

In an unpredictable environment like the one brought forward by the pandemic, early detection of clinical deterioration becomes a differentiating element in terms of quality. This, in turn, allows healthcare providers to review and adjust treatment before the patients gets worse. In the event it wasn’t possible to make enough changes, the provider can decide how to administer resources.

The solution

- In real time, a predictive model based on AI was developed. Said model can detect a change for the worse in the diagnosis of Covid 19 patients so as to assess the need of intensive care. Professionals gathered a sample of 1743 from those infected with Covid 19 who were hospitalized in three hospital centres in Spain. The model included variables according to current literature and applied data mining and machine learning techniques for its development. The predictive model was then generated with the dataset from February to December 2020 (1599 in total). It was made available to healthcare providers in September of 2020. The performance of the model was prospectively validated with the dataset of January 2021 – 144 in total

“Now we can generate alerts about the patients’ evolution, providing healthcare personnel the information they need to review the treatment before the patient’s condition gets worse. For us, this is very gratifying.”

Mireia Ladios
Product Owner de Cynara Care

Added Value

Its implementation has helped clinicians focus their attention on the most vulnerable patients. It also facilitates the decision-making process for managers with regards to limited resources and opportunity cost as is the care provided in the ICU.

1

Sensibility

0.98

Accuracy

0.99

AUC

0.98

Specificity

“It is a tool that allows us to anticipate and have the human resources and materials needed prepared to optimize the attention on patients in need of intensive care. It helps us, at an assistance level, to focus on those patients with higher risk so as to intensify our surveillance in the area of hospitalization.”

Manuel Mirón
Jefe de servicio de la UHD en el Hospital de Torrejón