The COVID-19 pandemic has transformed the landscape of healthcare with the exponential adoption of remote care and telehealth. On one hand, it exacerbated health disparities and on the other inspired the industry to create better off-the-shelf technologies for clinical care. The pandemic highlighted the need for a health delivery system efficient in daily life yet elastic enough to respond to unplanned shocks.

And AI has found early success in supercharging the healthcare operations and empower the entire health industry, from hospitals to public health departments. These successes are likely to only continue growing in the post-pandemic future and deepen the role of AI in digital health technology.

Drug Discovery

Drug development is an expensive and time-consuming process with clinical trials being the lengthiest phase. The existing computational approaches fall short of exploiting the available datasets of chemogenomics, pharmacogenomics, and side effects. AI, on the other hand, has demonstrated the potential to overcome these bottlenecks and augment drug research and development.

It offers an unprecedented opportunity to identify relevant protein and molecular targets (like COVID-19 spike protein), design candidate drugs, anticipate therapeutic responses and side-effects in patients, and thus optimize clinical trial enrollment. This can be a big push to the discovery, testing, and delivery of medicines in the future.

Telehealth 

Prior to the pandemic, only 1% of total doctor consultations took place virtually. However, the spike of COVID-19 ballooned this number to around 70% and highlighted the convenience and affordability of telemedicine. Besides, offering high-quality outcomes to doctors and patients, the adoption of digital health technologies has unfolded yet another possibility of AI augmentation.

The data generated from digital touchpoints like virtual appointments and chatbots create opportunities for enhancing remote healthcare delivery and offering more proactive, personalized care. From automating virtual patient visits and enabling remote clinical diagnosis from patients’ smartphone images to assessing the quality of text-based therapeutic services, the application of machine learning to mobile health data can enable real-time health analysis and personalized interventions.

Ambient Intelligence 

A lot of patients reach the ICU because of infections while thousands of others die because of medical errors. Artificial Intelligence can cut the rate of such fatal errors and help monitor vulnerable patients and make life-saving interventions. One such application can be AI sensors that would alert clinicians and visitors if they fail to sanitize their hands before visiting a patient. The sensors can flag hygiene failure by a colour transition.

Similarly, AI-enabled thermal sensors can be used for n visual tracking, human pose estimation, and activity recognition and cut down the repeated room-to-room surveillance by clinicians. The sensors can also detect changes habitual and behavioural changes in patients in daily living spaces where constant human observation is infeasible.

Medical Imaging

Hospitals and diagnostics face a lot of pressure to deliver diagnostic results rapidly and accurately, especially in times like COVID-19. While bigger hospital systems get jammed by the multitude of reports, the smaller clinics face a shortage of expertise. AI can help reduce this gap by assisting in advanced imaging and physician judgment. 

The application of AI to medical imaging promises dramatic progress in diagnostic tasks. AI-based models can analyze scans automatically and thereby enhance clinical accuracy and triage patients basis the urgency. They can also, enable a longer-range analysis of patients' imaging history, going beyond immediate clinical questions yet reducing costs.

Augmented Intelligence

Clinical judgment of patients with well-established diseases requires trials, literature, and research to understand disease prognosis and choose treatment options. However, the pandemic created a void in clinical care owing to the lack of clinical experience and decision-making. The resulting deferred treatments and exacerbated health problems have in turn incurred additional expenses on the healthcare system.

However, the AI-powered tools promise the potential to provide more personalized care and augment clinical decision-making by tapping existing patient records and genomic data. An example would be to develop a framework to predict the deterioration of patients with COVID-19 using a retrospective cohort of patients with similar respiratory diseases. 

As biomedical AI is swiftly moving into real-world use, addressing health disparities, it is imperative for all stakeholders - academia, industry, regulators, clinicians, patients, ethicists, etc. to ensure that it resolves these issues not exacerbate them. This falls in line with the vision of Fei-Fei Li, Professor of Computer Science and Co-Director of HAI,

“For AI algorithms, understanding dynamic human behaviour in health critical situations is far more challenging than labelling cats and chairs in images. From a technical perspective, many computer vision problems remain to be overcome before these technologies can be widely deployed.”