Investment and Business Opportunities created by Artificial Intelligence in Healthcare

The Artificial Intelligence market is rapidly growing, valued at $600 million in 2014 and projected to reach $150 Billion by 2026. The COVID-19 pandemic has led to the surge in adopting artificial intelligence and machine learning in pharmaceuticals and healthcare. Pharmaceutical companies and Industry stakeholders in healthcare acknowledge that AI has the answer to the most demanding challenges the healthcare system faces in present times. Artificial Intelligence has found its application in the drug development process, including selecting a Clinical Trial Site, executing hybrid clinical trials, and speeding up the process of bringing new drugs to the market. Apart from this, AI has multiple use cases throughout health plans, pharmacy benefit manager (PBM), Telemedicine, and many other health system enterprises. Software giants like Google, Nvidia, IBM, and many more are investing in Artificial Intelligence to open new business opportunities. Thus, AI has created business opportunities in risk stratification, genomics, imaging and diagnosis, precision medicine, and drug discovery. Coming to surgery, initiatives in Artificial Intelligence can create investment and business opportunities in the ecosystem of Computer-Assisted Surgery and Intervention, Image-Guided Surgery and Intervention, Hybrid Operating Room, and Guidance Systems. According to an article by Forbes, the economy of AI in surgery will reach $225.4 million in 2024. Today, business leaders in the Life Science and Healthcare Industry are progressing aggressively to make their systems more efficient with AI to provide top-notch patient care.

Business opportunities: Examples of startups in the field of artificial intelligence integrated surgery

Use of Artificial Intelligence Before, During, and After surgery

Pre-operative Phase

Artificial Intelligence in the preoperative phase is required to carry out accurate predicaments, automate and speed up the surgery planning procedure. When a patient is diagnosed with an illness that requires surgery to cure, the surgeon makes a preoperative plan based on the patient’s medical history and imaging tests to ensure that the surgery is carried out successfully. The preoperative phase uses general image-analysis techniques and traditional machine-learning boosted by deep-learning. Machine learning in this phase is used for anatomical classification, detection, segmentation, and image registration. Standard imaging tests used in practice include X-ray, CT-scan, ultrasound, PET, and MRI.

Some examples of the application of AI in the preoperative phase mentioned in the published article by Xiao-Yun Zhou et al. include:

• A proposal has been made to segment the lungs, bladder, and breast cancer types using the Convolutional Neural Network (CNN) architecture of Google’s Inception.
• Deep learning algorithms help recognize intracranial hemorrhage, calvarial fracture, midline shift, and mass effect during head CT scans.
• A deeply stacked autoencoder was trained to extract the statistical and kinetic biological features to detect prostate cancer from 4D Positron-Emission Tomography
• 3D CNNs (Convulsion Neural Networks) with roto-translation group convolutions were proposed to detect pulmonary nodules.
• Deep Reinforcement Learning (DRL) was used to detect breast lesions from dynamic contrast-enhanced MRI.
• Deep learning recurrent neural networks (RNN) have outperformed established clinical reference tools in predicting renal failure in real-time and mortality and postoperative bleeding after a cardiac surgery.

Intra-operative Phase

The intraoperative phase follows the preoperative phase, where the surgeon needs to perform the surgery. During this stage, the surgeon must be precise while making incisions or other surgical tasks. Performing surgery on the human body is a challenging task, and advancements in AI can assist the surgeon in facing challenging tasks during surgery. During the surgery, many surgical applications like ultrasound, NIRF (Near-Infrared Fluorescence), OCT (Optical Coherence Tomography), probe-based confocal laser endomicroscopy, electromagnetic sensors, and many other devices use AI techniques to provide computer-assisted surgery.

Recent works on AI for its use in intraoperative surgery can be divided into the following significant aspects:

• Intraoperative shape instantiation or 3D Shape instantiation-endoscopic Navigation, which includes depth estimation, visual odometry, and 3D reconstruction & localization
• Tissue Feature Tracking
• Augmented reality (superimpose images onto structures during open or minimally invasive surgeries.)

Use of robots during the intraoperative phase

Robotics surgery involves robots that help or collaborate with surgeons to perform minimally invasive surgeries and complex or repetitive procedures during surgery. Unlike traditional surgery tools, Surgical robots have the precision in controlling their trajectory, depth, and speed, and most importantly, they work without being fatigued. AI integrated robots can assist in reducing the impact of hand tremors and avoiding unintentional or accidental motions during surgeries, especially during surgery on the eyes. Robotics in healthcare has advanced to such an extent that surgeons can control the movement of abdominal surgical robots with the simple movement of eyes.

Artificial intelligence can give such surgical robots superhuman performance and utilize them to a great extent in minimal invasive surgeries (MIS). A few examples of the success of AI-assisted surgery specific to reproductive medicine includes

• AI-based robot-assisted minimally invasive transplantation of previously cryopreserved ovarian tissue with an extra-cellular tissue matrix scaffold, has shown success in the first pregnancies.
• AI has been beneficial in surgery conditions like vasectomy reversal in reproductive medicine.

AI in surgical robots boosts their systems in perceiving complex in vivo environments, conducting decision-making, and performing the desired tasks with increased precision, safety, and efficiency. The common AI techniques used for robotic and autonomous systems (RAS) can be summarized in four aspects:

1. Perception which involves surgical tools and environmental interaction, 3D tissues tracking & reconstruction, and instrument tracking & segmentation.
2. Localization and Mapping that involves simultaneous localization and mapping (SLAM), visual odometry & camera localization, and augmented reality.
3. System Modelling and Control which involves Kinematics and Dynamics Modelling, learning from demonstration, and reinforcement learning.
4. Human-Robot Interactions which involve cooperative control, touchless manipulation, intention understanding, and prediction

Some examples of innovations in AI-assisted surgery includes

• FAceMOUSe
• An AI-Driven microsurgery robot that sutures blood vessels in patients affected by lymphedema
• Robotic Hair Restoration
• Da Vinci cardio surgery which is robotic cardiac surgery.
• Gestonurse, which is a robotic scrub nurse that provides surgical instruments to surgeons in the operating room.

Post-operative Phase

The postoperative phase is the final phase of the surgery. It is the most intimate phase since the patient may be in pain, tired, or sleeping in an unaccustomed and sterile environment post-surgery. During this stage, the surgical team constantly looks for postoperative complications. The surgical team also attends to new patients and performs other surgeries lined in the department. Thus, this leads to increased stress, absenteeism, disengagement, and burnout among physicians. This presents a scientific and business opportunity to develop and implement such healthcare tools with AI that could predict postoperative complications and allow medical professionals to monitor high-risk patients more closely and effectively. One such example is electronic health records. Compared to clinical observation alone, natural language integration in electronic health records has enabled accurate detection of surgical site infections. Also, AI-integrated electronic health records are portable and automated. Currently, AI in electronic health records is still in the development stage.

Challenges in Artificial Intelligence Assisted Surgery

Challenges in the domain that have hampered the growth of Artificial Intelligence in surgery include:

• Difficulty in collecting substantial amounts of high-quality labelled training data.
• Difficulty in developing machine learning methods that are robust, accurate, and can deal with heterogeneous data sources.

• Challenges in collaboration with medical device manufacturers toward integrating devices, used in the surgery room with AI.
• Challenges in the post-operative care stage as described above.

Artificial Intelligence in surgery improves the ability of the surgeon to make better decisions on the need for acute surgery in a patient and accurately predicts the probability of success of any surgical procedure.