Scott Molina, IT Consultant03.26.20
It’s been said healthcare technology will change in the first quarter of the 21st century as much as communication technology changed from 1990 to 2015.
Over the past 20 years, medical technology has already seen considerable changes—for example, significant improvements to orthoscopic surgery and microsurgery, as well as the development of the wireless communications that concierge medicine relies on heavily. Now, rapid advancements are expected within the next half-decade in artificial intelligence (AI). In fact, they’ve already begun.
Following is a look at the most surprising and innovative ways AI will dramatically impact the healthcare industry.
1. Automated Epidemic Detection
On Dec. 31, 2019, BlueDot’s AI health monitoring system notified clients of an outbreak of what is now known as COVID-19—six days before the Centers for Disease Control and Prevention and nine days before the World Health Organization. The system first used language processing and machine learning algorithms to spot the incidence of the virus, then tracked airline ticket information to accurately predict where the illness would spread.
This is just a single example of how AI can “churn through” data faster and more accurately than even a team of expert data analysts. This success, paired with developing advancements, is likely to lead to further investment in and reliance on this kind of epidemic detection.
2. Crowdsourcing Mental Health
As a concept, asking for mental health advice from friends and trusted advisors has been happening for eons. But it took on more life and scale during the internet age through chat rooms for personal advice, online support groups, and later, apps to help with self-diagnosis and self-care.
AI is making a play at this level with Mental Health America’s diagnostic tests for depression, anxiety, PTSD, and addiction. As technology develops, we can expect to see AI-driven tools for therapy (especially cognitive behavioral therapy and similar techniques for regulating real-time behavior), diagnosis, and emergency intervention.
3. Faster Suicide Prevention
In addition to being a diagnostic tool for mental health, AI’s potential role in suicide prevention is exciting. Specifically, AI can mine social media posts, text messages, emails, and similar platforms to search for key phrases regarding self-harm. It uses the same techniques Facebook and Google do to populate your advertisements.
Ultimately, researchers could pair these tools with AI-driven phone apps and therapy apps to provide a holistic solution to suicidal tendencies and self-harm impulses, providing the right help at the right time.
4. Improving Medical Imaging
It’s not entirely accurate to say AI will change medical imaging in the next five years. Imaging has already been disrupted by AI, and there will be even more changes to come.
Deep learning and neural networks, applied to radiology and other imaging techniques, have a profound potential for making medical imaging faster and more accurate. In a 2018 workshop at the National Institutes for Health in Bethesda, Maryland researchers demonstrated theoretical frameworks and prototypes for AI assistance across image reconstruction, noise reduction, triage, and detection.
5. Identifying Domestic Violence Risk
Any ER employee or law enforcement officer knows underreporting is one of the most frustrating and dangerous challenges in stopping domestic violence. Although an exact number is impossible to determine, expert estimates say about 50 percent of domestic violence is unreported.
AI programs built to identify patterns of injury and behavior can be paired with ER medical imaging to identify signs of domestic violence a human worker might miss due to subtlety or simple overwork. The program would then flag the case for exploratory conversation and follow-up investigation as needed.
Although AI-supported programs have yet to see broad application, initial research on predicting domestic violence has been promising.
6. Automated Stroke Support
Seconds count during stroke treatment, with each moment that elapses equating to greater brain damage, more debilitation, and a higher likelihood of death. In areas far from a metropolitan medical center, those seconds add up quickly. AI assists with this in two significant ways.
First, it can close the gaps in access to imaging analysis and expert diagnosis between hospitals in developed areas and those in developing nations or rural areas. That can shave minutes off the time required to locate a clot and begin repairs.
Second, ongoing research is investigating how to provide on-site, automated, AI-driven tools to help patients recognize a stroke at its onset, or to automate its detection through biometrics checked via phone app or portable device. For instance, people currently wear pacemakers for heart issues, but similar devices could be used for individuals at risk of a stroke.
7. Augmented Diagnostics
The systems-changing "Checklist Manifesto" wrought sweeping disruption in many healthcare industries by encouraging diagnostic checklists, especially in emergency rooms and surgical care. AI is poised to capitalize on this because properly designed computers are simply better at running through checklists than humans.
When in place, a neural network program can access biometrics, imaging, and questionnaire answers to offer an initial diagnosis in seconds, rather than the minutes (or hours, including wait time) currently experienced in some emergency cases.
8. Faster, Confidential Information Exchange
Combined with the Fast Healthcare Interoperability Resources system, AI has the potential to improve both treatment outcomes and financial viability for healthcare across the world. Sharing confidential information about a patient is time-consuming because of the proper permissions necessary. That added time can cause worse outcomes and leads to higher administrative costs, which impact overall patient costs.
Applying AI to the data-mining aspects of this task means faster discovery and reporting of the most important patient information. Further, because an algorithm is the only thing looking at full medical records, there’s the potential to share information more swiftly and with less red tape.
Although this strategy hasn’t gained widespread acceptance, a 2019 roundtable report by the Center for Open Data Enterprise recommends and predicts five immediate applications for existing AI technology:
9. More Equal Access to Specialized Care
The internet and IT revolution have already made great leaps in this area through innovations ranging from telediagnostics to remote surgery, allowing a specialist located in a metropolis to help professionals in rural areas. This provides better outcomes for those in developing areas.
AI stands to capitalize on this initial advance by being “present” in multiple areas at once. A well-developed algorithm can perform expert diagnostics anywhere it’s installed, at a fraction of the price required to hire a human specialist in the area. For example, a rural hospital without a stroke care specialist can instead use an installed AI algorithm to check imaging and symptoms to make a diagnosis more quickly and accurately than calling in a qualified doctor for a video consultation.
Scott Molina lives in Boston. He’s an IT consultant who works with the healthcare industry, as well as financial services firms.
Over the past 20 years, medical technology has already seen considerable changes—for example, significant improvements to orthoscopic surgery and microsurgery, as well as the development of the wireless communications that concierge medicine relies on heavily. Now, rapid advancements are expected within the next half-decade in artificial intelligence (AI). In fact, they’ve already begun.
Following is a look at the most surprising and innovative ways AI will dramatically impact the healthcare industry.
1. Automated Epidemic Detection
On Dec. 31, 2019, BlueDot’s AI health monitoring system notified clients of an outbreak of what is now known as COVID-19—six days before the Centers for Disease Control and Prevention and nine days before the World Health Organization. The system first used language processing and machine learning algorithms to spot the incidence of the virus, then tracked airline ticket information to accurately predict where the illness would spread.
This is just a single example of how AI can “churn through” data faster and more accurately than even a team of expert data analysts. This success, paired with developing advancements, is likely to lead to further investment in and reliance on this kind of epidemic detection.
2. Crowdsourcing Mental Health
As a concept, asking for mental health advice from friends and trusted advisors has been happening for eons. But it took on more life and scale during the internet age through chat rooms for personal advice, online support groups, and later, apps to help with self-diagnosis and self-care.
AI is making a play at this level with Mental Health America’s diagnostic tests for depression, anxiety, PTSD, and addiction. As technology develops, we can expect to see AI-driven tools for therapy (especially cognitive behavioral therapy and similar techniques for regulating real-time behavior), diagnosis, and emergency intervention.
3. Faster Suicide Prevention
In addition to being a diagnostic tool for mental health, AI’s potential role in suicide prevention is exciting. Specifically, AI can mine social media posts, text messages, emails, and similar platforms to search for key phrases regarding self-harm. It uses the same techniques Facebook and Google do to populate your advertisements.
Ultimately, researchers could pair these tools with AI-driven phone apps and therapy apps to provide a holistic solution to suicidal tendencies and self-harm impulses, providing the right help at the right time.
4. Improving Medical Imaging
It’s not entirely accurate to say AI will change medical imaging in the next five years. Imaging has already been disrupted by AI, and there will be even more changes to come.
Deep learning and neural networks, applied to radiology and other imaging techniques, have a profound potential for making medical imaging faster and more accurate. In a 2018 workshop at the National Institutes for Health in Bethesda, Maryland researchers demonstrated theoretical frameworks and prototypes for AI assistance across image reconstruction, noise reduction, triage, and detection.
5. Identifying Domestic Violence Risk
Any ER employee or law enforcement officer knows underreporting is one of the most frustrating and dangerous challenges in stopping domestic violence. Although an exact number is impossible to determine, expert estimates say about 50 percent of domestic violence is unreported.
AI programs built to identify patterns of injury and behavior can be paired with ER medical imaging to identify signs of domestic violence a human worker might miss due to subtlety or simple overwork. The program would then flag the case for exploratory conversation and follow-up investigation as needed.
Although AI-supported programs have yet to see broad application, initial research on predicting domestic violence has been promising.
6. Automated Stroke Support
Seconds count during stroke treatment, with each moment that elapses equating to greater brain damage, more debilitation, and a higher likelihood of death. In areas far from a metropolitan medical center, those seconds add up quickly. AI assists with this in two significant ways.
First, it can close the gaps in access to imaging analysis and expert diagnosis between hospitals in developed areas and those in developing nations or rural areas. That can shave minutes off the time required to locate a clot and begin repairs.
Second, ongoing research is investigating how to provide on-site, automated, AI-driven tools to help patients recognize a stroke at its onset, or to automate its detection through biometrics checked via phone app or portable device. For instance, people currently wear pacemakers for heart issues, but similar devices could be used for individuals at risk of a stroke.
7. Augmented Diagnostics
The systems-changing "Checklist Manifesto" wrought sweeping disruption in many healthcare industries by encouraging diagnostic checklists, especially in emergency rooms and surgical care. AI is poised to capitalize on this because properly designed computers are simply better at running through checklists than humans.
When in place, a neural network program can access biometrics, imaging, and questionnaire answers to offer an initial diagnosis in seconds, rather than the minutes (or hours, including wait time) currently experienced in some emergency cases.
8. Faster, Confidential Information Exchange
Combined with the Fast Healthcare Interoperability Resources system, AI has the potential to improve both treatment outcomes and financial viability for healthcare across the world. Sharing confidential information about a patient is time-consuming because of the proper permissions necessary. That added time can cause worse outcomes and leads to higher administrative costs, which impact overall patient costs.
Applying AI to the data-mining aspects of this task means faster discovery and reporting of the most important patient information. Further, because an algorithm is the only thing looking at full medical records, there’s the potential to share information more swiftly and with less red tape.
Although this strategy hasn’t gained widespread acceptance, a 2019 roundtable report by the Center for Open Data Enterprise recommends and predicts five immediate applications for existing AI technology:
- Increasing both access to and privacy protection for patient data
- Unified standards of data quality
- Removing barriers to data sharing between care organizations
- Clarifying the appropriate use of patient information
- Reducing bias while increasing accountability
9. More Equal Access to Specialized Care
The internet and IT revolution have already made great leaps in this area through innovations ranging from telediagnostics to remote surgery, allowing a specialist located in a metropolis to help professionals in rural areas. This provides better outcomes for those in developing areas.
AI stands to capitalize on this initial advance by being “present” in multiple areas at once. A well-developed algorithm can perform expert diagnostics anywhere it’s installed, at a fraction of the price required to hire a human specialist in the area. For example, a rural hospital without a stroke care specialist can instead use an installed AI algorithm to check imaging and symptoms to make a diagnosis more quickly and accurately than calling in a qualified doctor for a video consultation.
Scott Molina lives in Boston. He’s an IT consultant who works with the healthcare industry, as well as financial services firms.