Pooja Rao
Smart intelligence for trauma caregivers
May 17, 2019
By Pooja Rao
Artificial intelligence (AI) has been transforming the world in many ways, and the field of medicine is no exception. It has been invaluable in aiding the diagnosis and treatment of many life-threatening conditions. In fact, the research firm CB Insights reports that AI-driven healthcare startups had raised $4.3B across 576 deals between 2013 and 2018, overtaking all other industries in AI deal activity.
The use of AI and machine learning technologies for healthcare is not a new concept. FDA-approved algorithms were used as early as 1998, for detecting cancerous cells. The key difference today is that the hardware and the algorithms have improved substantially, which enables us to deliver faster, innovative and accessible healthcare solutions. AI has now been deployed for a wide range of medical needs, from streamlined workflows to robot-assisted surgeries
Why you should know about TBI
March was Traumatic Brain Injury (TBI) awareness month. TBI occurs due to sudden and excessive force on the skull, and being aware is critical as it could happen to anyone, at any time. In the US, around 137 people die from a TBI every day. From bad falls and vehicle collisions to violent assaults, or even tumors, TBIs are incredibly dangerous. Children are often the most susceptible, and the effects of such an injury to their developing brains can, in many cases, become permanent.
Patients who’ve had serious head injuries may face permanent changes in personality, physical abilities and thinking processes. This is because TBI patients’ brains show decreased activity in the prefrontal cortex, temporal lobes and cerebellum – the regions that govern self-control of mood and behavior, memory formation and coordinated.
A head injury can occur even when there are no visual signs of injury on the head itself, and therefore it can sometimes be hard to assess its level of seriousness. The head has more blood vessels than any other part of the body, so any bleeding on the brain’s surface or within, is a serious concern. Some minor head injuries bleed a lot, while some major injuries don’t bleed at all, therefore it’s critical all head injuries be assessed by a doctor and treated immediately. TBI diagnosis in an emergency unit is usually done through a CT scan, which enables the doctor to assess the extent of the injury and determine the need for surgery. A detailed MR scan is followed up once the patient is stabilized.
AI to the rescue
Studies have proved that TBI patients’ recovery depends heavily on the speed to initiate treatment. Patients treated within 90 minutes of the injury have a better chance of improving within 24 hours and have better outcomes within a 3-month timeframe.
However, the time sensitive nature of these injuries is compounded by the constant and drastic shortage of experienced radiologists. Their efficacy could be highly improved by an automated prioritization of incoming cases, that relies on improving patient mortality and outcomes.
TBI injuries are often too critical to be left to trainee radiologists. Even if they conduct the CT scan in time, an inaccurate measurement could lead to wrong treatment choices which would directly affect the patient’s survival and recovery. It also opens up the possibility of legal action against the radiologist and the healthcare provider.
AI has played a crucial role in accurately and speedily identifying TBI patients with critical abnormalities, as well as quantifying and localizing the bleeding in their brains. It then combines these findings to prioritize the cases that need immediate physician attention.
Some AI solutions also display the detected abnormality overlaid on the original CT scan image, so that the radiologist can quickly verify the accuracy of the AI-powered decision. The radiologist can then swiftly browse across multiple CT scans.
Pointers that will aid in selecting a smart AI solution:
Extent of global deployments the solution provider has is a statement to their success.
The number of datasets on which the algorithm has been trained. Higher the dataset, better the solution.
Number of peer reviews the solution provider has with reputed research firms add to their credibility
Always validate the solution on your own datasets before implementing.
Using AI in such time sensitive scenarios demonstrates the value intelligent neuroimaging offers to precision medicine, targeted care and the potential for improved outcomes.
About the author: Pooja Rao is the co-founder and R&D head of Qure.ai.
Artificial intelligence (AI) has been transforming the world in many ways, and the field of medicine is no exception. It has been invaluable in aiding the diagnosis and treatment of many life-threatening conditions. In fact, the research firm CB Insights reports that AI-driven healthcare startups had raised $4.3B across 576 deals between 2013 and 2018, overtaking all other industries in AI deal activity.
The use of AI and machine learning technologies for healthcare is not a new concept. FDA-approved algorithms were used as early as 1998, for detecting cancerous cells. The key difference today is that the hardware and the algorithms have improved substantially, which enables us to deliver faster, innovative and accessible healthcare solutions. AI has now been deployed for a wide range of medical needs, from streamlined workflows to robot-assisted surgeries
Why you should know about TBI
March was Traumatic Brain Injury (TBI) awareness month. TBI occurs due to sudden and excessive force on the skull, and being aware is critical as it could happen to anyone, at any time. In the US, around 137 people die from a TBI every day. From bad falls and vehicle collisions to violent assaults, or even tumors, TBIs are incredibly dangerous. Children are often the most susceptible, and the effects of such an injury to their developing brains can, in many cases, become permanent.
Patients who’ve had serious head injuries may face permanent changes in personality, physical abilities and thinking processes. This is because TBI patients’ brains show decreased activity in the prefrontal cortex, temporal lobes and cerebellum – the regions that govern self-control of mood and behavior, memory formation and coordinated.
A head injury can occur even when there are no visual signs of injury on the head itself, and therefore it can sometimes be hard to assess its level of seriousness. The head has more blood vessels than any other part of the body, so any bleeding on the brain’s surface or within, is a serious concern. Some minor head injuries bleed a lot, while some major injuries don’t bleed at all, therefore it’s critical all head injuries be assessed by a doctor and treated immediately. TBI diagnosis in an emergency unit is usually done through a CT scan, which enables the doctor to assess the extent of the injury and determine the need for surgery. A detailed MR scan is followed up once the patient is stabilized.
AI to the rescue
Studies have proved that TBI patients’ recovery depends heavily on the speed to initiate treatment. Patients treated within 90 minutes of the injury have a better chance of improving within 24 hours and have better outcomes within a 3-month timeframe.
However, the time sensitive nature of these injuries is compounded by the constant and drastic shortage of experienced radiologists. Their efficacy could be highly improved by an automated prioritization of incoming cases, that relies on improving patient mortality and outcomes.
TBI injuries are often too critical to be left to trainee radiologists. Even if they conduct the CT scan in time, an inaccurate measurement could lead to wrong treatment choices which would directly affect the patient’s survival and recovery. It also opens up the possibility of legal action against the radiologist and the healthcare provider.
AI has played a crucial role in accurately and speedily identifying TBI patients with critical abnormalities, as well as quantifying and localizing the bleeding in their brains. It then combines these findings to prioritize the cases that need immediate physician attention.
Some AI solutions also display the detected abnormality overlaid on the original CT scan image, so that the radiologist can quickly verify the accuracy of the AI-powered decision. The radiologist can then swiftly browse across multiple CT scans.
Pointers that will aid in selecting a smart AI solution:
Extent of global deployments the solution provider has is a statement to their success.
The number of datasets on which the algorithm has been trained. Higher the dataset, better the solution.
Number of peer reviews the solution provider has with reputed research firms add to their credibility
Always validate the solution on your own datasets before implementing.
Using AI in such time sensitive scenarios demonstrates the value intelligent neuroimaging offers to precision medicine, targeted care and the potential for improved outcomes.
About the author: Pooja Rao is the co-founder and R&D head of Qure.ai.