Implementing Artificial Intelligence for effective health management
Published: Sun 3 Jan 2021, 10:07 AM
Last updated: Sun 3 Jan 2021, 12:50 PM
American Hospital Dubai, part of Mohamed & Obaid Al Mulla Group, in partnership with global healthcare technology company Cerner, has announced the first outcomes of one of the models from its artificial intelligence (AI) data lab. The lab serves as a prototype in the region for developing clinically-driven AI techniques with proven predictive model outcomes to serve the healthcare sector.
Sherif Beshara, chief executive officer of Mohamed & Obaid Al Mulla Group, said: "American Hospital Dubai is one of few leading organisations in this region who have proactively taken the lead to invest in AI with the goal of enhancing patient and caregiver experiences. Together with our key strategic partners such as Cerner, we share the UAE Vision 2021 and align with UAE's National Program that identifies AI as one of the key strategies to help transform healthcare in the country."
American hospital Dubai has embarked on a multi-phase AI transformation programme, with the first phase being a data lab built within American Hospital Dubai's datacentre. The lab's data scientists leverage advanced technologies and Cerner's intelligence predictive models.
The first of four models planned by American Hospital Dubai for the AI transformation programme is Covid-19 Patient Mortality and ICU Admission Prediction. This model aims to have risk identifiers available to identify patients who are at high risk for ICU admissions and risk of mortality; use predictors and Cerner intelligence to stratify the population at risk and provide timely and appropriate care; and reduce the ICU admission and Covid-19 related mortality.
American Hospital Dubai has used multiple approaches for this first model: 50 risk identifiers across the globe have been considered and more than 17,000 patient records have been analysed, so as to identify the co-relations of clinical, demographical and comorbidity data.
American Hospital Dubai has also begun investigating large datasets of Covid-19 patients with a view to predicting higher risk of mortality and morbidity. Fifty different parameters were formulated into a Covid-19 Patient Mortality and ICU Admission Prediction model and fed to the machine learning algorithms. Some of the initial outcomes over the past few months are :
- Quantification of correlation between clinical outcomes and existing comorbidities. An investigation showed that the top three chronic conditions that contribute to higher ICU admissions and mortality rate were hypertension (34 per cent), diabetes (22 per cent) and chronic kidney disease (22 per cent). Around 60 per cent of the patients admitted to ICU have obesity as a comorbid condition. All patients flagged with these conditions will follow a specific care pathway.
- Pattern recognition for personalised approach. The risk of ICU admissions is 50 per cent higher for female patients having a body mass index above 25 when they tested positive for Covid-19.
- Identification of key clinical indicators. A study showed that high creatinine levels and high white blood cells were identified as important indicators for patients who would be admitted to ICU, while albumin was identified as the single most significant contributor for predicting those patients who would not require ICU admissions with an accuracy of 80 per cent.
Beshara stated: "The American Hospital Dubai AI transformation programme and the data lab are fulfilling one of our key objectives of leveraging healthcare innovative tools to proactively care for individuals and help reduce chronic diseases, thereby having a healthier population positively influenced to own and manage their health."
Other models currently being built by American Hospital Dubai are predicting which asthma patients have the greatest risk of ED admissions, forecasting appointment no-shows and projecting ED hospital admissions. The outcomes are expected to be available after two months. American Hospital Dubai has also developed an onboarding framework for other UAE facilities and clinics that are keen to contribute to this programme and feedback meaningful outcomes to drive clinical, operational and financial efficiencies.
The next phase of the programme, once the remaining models have been built, localised and implemented at American Hospital Dubai's data lab, will be for American Hospital Dubai to leverage world-class AI services and an intelligent data science environment. The environment will host Cerner's intelligence tools, such as HealtheDataLab and HealtheIntent, which are key for transitioning to the next stage of executing the AI program on a wider healthcare industry level.
"AI has a key role to play in advancing healthcare innovations given the increase of big data, the growing adoption of precision medicine, enhancements in computing power and the heightened focus on quality of patient care. We congratulate American Hospital Dubai and its team on their achievements with the first model over the past few months and are honoured to continue supporting them throughout their AI transformation journey," said Alaa Adel, managing director, Cerner Middle East and Africa.