Advanced search
Start date

SADIAX, an innovative computer-vision-based solution for monitoring patients in intensive care units (ICU)


According to data from the National Registry of Health Establishments (CNES), Brazil has almost 41,000 Intensive Care Unit (ICU) beds. Half of them are allocated to serve approximately 204 million Brazilians through of the Unified Health System, while the others half are used on the private health care system to serve up to 50 million. In São Paulo, the number of ICU beds corresponds to 30 % of the total amount in Brazil. In the United States, more than 5.7 million adults are admitted to the ICU annually, costing the health sector more than $ 67 billion per year. The quality of hospitalization and the patient's length of stay in the ICUs are essential factors for both patients and health institutions. The monitoring of patients in ICU beds is a deterministic activity in the evaluation of these factors. When not well monitored, patients may have an extended stay, increasing their risk of mortality. Commonly, monitoring systems are used to continuously check the patient's vital signs to avoid any critical disturbance, such as agitation, sedation, and delirium. Those disturbances can be measured by using subjective methods. Thus, a confidence identification of those problems is a great challenge to be overcome. Pain is a widespread symptom in ICU beds, and its late stages make it difficult to properly treat the symptom, causing critical damage to the patient. Thus, it is essential that it be detected quickly on its early stages. The monitoring of agitation and sedation levels is also a vital issue for the safety of patients, given the fact that when they wake up from a sedation process patients tend to be agitated, increasing the likelihood of falls and the removal of respiration masks and accesses, which can be fatal. A poor evaluation or non-detection of these disturbances, due to environmental factors, such as excessive light and noise, can generate a common risk factor in hospitals: the delirium. The occurrences of delirium are much common on adults and elders and, if not correctly detected, its consequence can be fatal. In order to deal with those issues, various researches and new technologies have already been carried out using body sensors and cameras. However, lack of precision may prevent these technologies from being commercially exploited. The SadiaX project has been developed and validated by the company since January 2018 at the Israelita Albert Einstein Hospital and the Johnson & Johnson Innovation Lab in Houston, TX, where a company is also incubated. Evidence, the data on the prototype of SadiaX, the first system of monitoring of patients in the UTI to detect of 10 levels of pain, agitation, sedation and delirium. During these last months, we developed the initial technology and the prototype of SadiaX, the first ICU bedside patient monitoring system to detect 10 levels of pain (discomfort), agitation, sedation and delirium. From our validation results and our current prototype, the methodologies for the next steps of the project are: i) combine the functionalities of SadiaX into a concise platform, ii) research and refine technologies, iii) develop the agitation and sedation detector ,iii) develop the delirium-propensity detector, iv) develop as user-centered interfaces, all three types of users, v) develop a cloud platform that aggregated information during patient monitoring and, vi) perform a clinical validation. The main result will be a kit consisting of software and hardware that can be installed in ICU beds to accurately detect levels of pain, agitation, sedation, and delirium. The system aims to increase the health professional's control over the patient's condition, improving the level of comfort and quality of hospitalization. Thus, it will reduce the number of falls in bed caused by agitation, and also the system will improve the delirium diagnosis. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items

Please report errors in scientific publications list by writing to: