The video addresses the pressing issue of health complications due to falls among seniors and the broader implications on healthcare. With an aging population, falls present not only tragic personal consequences but also significant financial burdens, as costs for post-fall care are projected to rise drastically in the coming years. The video highlights the unpreparedness of the healthcare system, particularly with an insufficient number of providers like nurses and physicians, further exacerbated by the COVID pandemic, leading to a critical shortage and burnout among existing staff.

The discussion then shifts to the potential of Artificial Intelligence (AI) in transforming healthcare delivery. AI is presented as a solution to many current challenges, emphasizing its capacity for early detection, personalization of care, and continuous monitoring, which could prevent falls and other health conditions before they escalate. The integration of AI offers the opportunity for more efficient, data-driven healthcare by using both structured and unstructured data, breaking down current care silos, and providing early warning systems for at-risk patients.

Main takeaways from the video:

Please remember to turn on the CC button to view the subtitles.

Key Vocabularies and Common Phrases:

1. fall risk [fɔːl rɪsk] - (noun) - Potential likelihood of a person, particularly a senior, to fall. - Synonyms: (hazard, peril, danger)

She was never diagnosed as a fall risk.

2. longitudinal history [ˌlɒndʒɪˈtjuːdɪnəl ˈhɪstəri] - (noun) - Record of an individual's medical information over a long period. - Synonyms: (timeline, chronicle, annals)

We have longitudinal history on various people across decades now.

3. polypharmacy [ˌpɒlɪˈfɑːməsi] - (noun) - The use of multiple medications by a patient, particularly when more than are medically necessary. - Synonyms: (multi-medication, overmedication, drug interaction)

The secondary conditions are polypharmacy, things like drug reactions.

4. unstructured data [ʌnˈstrʌktʃə(r)d ˈdeɪtə] - (noun) - Information that is not organized in a pre-defined manner. - Synonyms: (chaotic data, non-sequential data, irregular data)

And there is a lot of it as unstructured data where you scan a document and put it in there.

5. gene sequencing [ʤiːn ˈsiːkwənsɪŋ] - (noun) - Process of determining the order of the nucleotides in DNA. - Synonyms: (DNA mapping, genetic sequencing, genome analysis)

This is going to be on boosters in the next five years. gene sequencing, understanding what mutations actually lead.

6. precision medicine [prɪˈsɪʒən ˈmɛdɪsɪn] - (noun) - Medical approach that tailors treatment to individuals based on their genetic content or other molecular or cellular analysis. - Synonyms: (personalized medicine, tailored treatment, individualized therapy)

There are things like precision medicine where once you identify, you can design a specific medicine.

7. crispr [ˈkrɪspər] - (noun) - A gene-editing tool enabling alterations in DNA sequences. - Synonyms: (gene editing, DNA modification, genetic engineering)

There are things like crispr, which is a gene editing tool before people get heart diseases or cancer.

8. demographic [ˌdɛməˈɡræfɪk] - (noun) - A specific segment of the population distinguished by common characteristics such as age, race, or gender. - Synonyms: (population group, cohort, census category)

And most of the studies are conducted on white males and females. And the other populations are not appropriately reflected today.

9. comorbidities [kəʊˌmɔːˈbɪdɪtiz] - (noun) - Simultaneous presence of two or more diseases or medical conditions in a patient. - Synonyms: (concurrent conditions, coexisting conditions, multiple illnesses)

And the systems can actually tell you what is a better drug for a patient that would work best given the primary and the secondary diagnosis and the comorbidities they have.

10. trifecta [traɪˈfɛktə] - (noun) - A situation involving three elements, especially three conditions or factors coming together to affect an outcome. - Synonyms: (trinity, triplet, triumvirate)

There is a trifecta coming together with aging population, no caregivers, and the cost going out of control.

How AI could facilitate preventative healthcare - Deepak Gaddipati - TEDxBradley University

Good evening. How many of you in the room have a loved one that experienced a fall? It is a very common problem. One in four seniors fall every year. My grandmother, who was 68 years old, on her way to the bank fell, broke her hip and died within 10 days from the day she fell. She never fell before. That was her first fall. She was never diagnosed as a fall risk. Post fall, seniors see a dramatic decline in quality of their lives. Over 30% of seniors who have a fall die within the first 12 months. 50%. If you look at the US economy, we are spending over $50 billion on falls after a fall happens, which is a lot of money.

And if you think about it, it's not just the falls, it's about other preventable conditions like pressure injuries. Every year 50,000 people die and it costs us $10 billion to take care of people with pressure injuries. If you think about infections, over 300,000 people die and it cost us about $30 billion. Between all these three conditions I just mentioned, falls, pressure injuries and infections, we're spending over $90 billion and over 10 million people get affected. And this doesn't have to be this way. Falls and pressure injuries, all of these are preventable.

And if you look at what is going ON in the U.S. today, about 17% of the population are seniors who are 65 and older. In the next few years, by end of 2030, 21% of our population will be seniors. One in every five people. We are an aging country and we need more providers to take care of our aging communities and population. We need more nurses, we need more physicians and we need more care coordinators. American Nurses association recently put out a study which is really unfortunate that we are short of about 1.2 million nurses in United States and over 40,000 physicians in the US. These are mind bogglingly, staggeringly bad stats.

And if you think about it, what is going to happen to all these seniors who are aging now? Before COVID this was a trend of declining clinical providers in the country. But post Covid, a lot of clinicians had sickness because of COVID They carried it home, their kids got sick, their family members who are immunocompromised or having other health conditions, they got sick and it caused such bigger problem in their families that they decided to quit their professions in healthcare and become someone else, something else. I've known personally a bunch of people who became real estate agents, who were clinicians before and this is truly an unfortunate trend.

So what is happening now? So the nurses and the physicians who are in the hospitals and nursing homes, they have more patients. The nurse to patient ratio is up the roof at record levels in the country today. The physician to the patient ratios are out of bounds right now. And there is stress, long working hours and burnout. And this is real. So we already have aging population shortage and current nurses are getting stressed out and physicians are burnt out. How do we deal with this? These are really tough problems that are not going to go away by themselves.

And if you look on the other side, what is happening to our costs? 2012 to insure a family, private insurance costed about $16,000 a year. Today that number is about $24,000. So which is 50% higher? These numbers are mind bogglingly high. These are growing at over 5% year over year. And this is not sustainable. And how are we going to take care of all these people who are getting old? This is, there is a trifecta coming together with aging population, no caregivers, and the cost going out of control. And this needs to be fixed. I'll take a very simple problem that I started with which is falls.

If you look at falls in 2012, it used to cost the US economy about $50 billion a year to provide care after someone falls. And today that number is about $70 billion. And very soon, by end of 2030, that number is going to become over $100 billion. And what do we do about that today? Today, when seniors go for their annual wellness visits, the physicians ask, have you fallen? How many times have you fallen and was it a fall with injury? And they take all the data and put it in the Ehrlich and that's the end of it. Nothing happens with it. There is no, if you look at the healthcare today, we take care of people when they get sick. We are not a true healthcare because we only worry when someone is really experiencing problems.

So that has to change. We are spending billions if not trillions of dollars on problems that could have been fixed in the first place. So if you look at healthcare, care is delivered in four different silos today. You have care delivered in home, you have care delivered in ambulatory locations like physicians, primary care, neurology, orthopedics, you have care delivered in hospitals and you have care delivered in post acute care, which is skilled nursing and memory care. So these four data silos, I mean there is various softwares, various EHRs and various pieces of information sitting in various places today.

So some of it is structured data. What is structured data? You know, when there is a name, you put in a name As X, Y, Z and the first name, last name and date of birth. So you're putting it in specified fields, I.e. structured data and there is a lot of it as unstructured data where you scan a document and put it in there. No one is ever going to look at the document. So the good point that happened in the last two decades is EHRs became a prominent thing. We have longitudinal history on various people across decades now, but no one is actually looking and trying to figure out what is making sense of this data yet.

We are using human driven statistical models to go pull the data and see what are very broad trends and how do we address the broad trends. And that's what we are doing today with what we have. With the advent of AI, we can actually use data driven insights to actually look at all this data across silos, structured and unstructured and see a patient as a patient, not as a identification number in one EHR and connect all the dots. And that is going to really transform the way care is delivered because now you have a continuous history of this person, you just don't have a very specific two day visit history that you're basing all your care on.

So that is going to be the first big benefit that AI models are going to provide. And we are seeing that today in a lot of health systems they're using models to do this. The second benefit comes from personalization understanding. Today a physician or some clinician has to look at the data and make sense and give you a recommendation on what to do. So it might be a very simple thing. You went and got your labs done and you can't see the data until the physician tells and interprets it for you what it means.

So that is changing. These days you're able to automatically analyze the data and provide that analysis on fingertips for patients so they can immediately know what is going on with them. And if they have any questions they can ask AI agents that can actually explain it to them on what is going on. These are not humans, these are all software programs that are able to do that. And there is quite a few health systems that is all testing and doing a bunch of these things. And the third part, early detection, Knowing before something goes wrong, monitoring people continuously to figure out what is going on with patients when they are not in the hospital, when they are in the hospital 24, seven, not once every two hours, once every second, or ten times every second.

So this much data today, everything is going to a set of two, three people and they have to make a decision and everyone else is waiting for the decision to be made. So it's a waiting game when you're in the house or when you're in the hospital and all these things will start changing. You will see things move so drastically fast. Now you don't need a person to call and tell you there's an appointment available in two weeks. This is going to automatically happen. You call in, they'll get you in the next best appointment. There's no human factor involved. Small questions will automatically be answered.

So again, coming back to the case study about false the technology you see here from a false perspective does all these three things. So if you go and ask today what is the primary diagnosis that leads to false? Its neurological issues, musculoskeletal issues. So these are the neurological and muscular conditions cause lot of fall related diagnosis for people. And if you look at it how this is actually, what are the secondary conditions that cause it? Everyone knows the primary condition, but no one knows the secondary conditions.

The secondary conditions are polypharmacy, things like drug reactions that happen that the patient started changing, dehydration, they didn't take enough water. And there is a slew of other factors. All these are coming out because AI algorithms are able to go in and figure these out ahead of time. And on top of that, personalization. So the technology you see in the picture here uses a technology called lidar and it can figure out how do people walk. It can look at their gait speed, the speed at which they walk, their step length, there's stride length, and it can look at all these from that person's perspective and tell what is the likelihood that someone is going to fall in the next 12 months and what you need to do to actually provide care to minimize the risk for them to fall.

So the clinicians can design a care plan and reevaluate the patient and see how better or worse they're doing. That's what is the possibility that we can do with the AI coming to the benefits. There's a lot of benefits with AI. Access to care immediately. If you want something, the knowledge is on your fingertips that are customized for you. Digital coaches, the third biggest beneficiary I see is human AI interaction. So today if you think about physicians are trained on millions of people data and they don't know some very minute educated scenarios. And it's very hard to think about them because they see these cases once in a lifetime, if in their lifetime.

So what AI is able to do is work with the clinicians to flag these situations. So they can actually think about these way ahead of time and alert the patients. So these are things that truly will benefit us in terms of using AI. And there are some ethical considerations. AI is designed on what is data that is available. And most of the studies are conducted on white males and females. And the other populations are not appropriately reflected today. If you think about it, the average age of a white person in United states is about 78 years. For an Asian person it's about 86 years.

For an African American person, it's 72 years. You can't take a curve that is based on a specific age group and fit for every broad population. So we need to have an open dialogue and figure out how we are going to do this. So this is all I spoke about is what is happening today, actually the next gen of AI. This is going to be on boosters in the next five years. gene sequencing, understanding what mutations actually lead based on the genome of the person and figuring out as an example, would Tylenol be a better drug for this patient or would ibuprofen be a better drug for the patient?

So they will understand the genome, the primary and the secondary diagnosis and the comorbidities they have. And the systems can actually tell you what is a better drug for a patient that would work. Deep learning, machine learning models can look at all the patterns of genome sequencing over time and see what is a specific cancer causing genome link that need that can be identified way ahead of time across this population pool. There are things like precision medicine where once you identify, you can design a specific medicine that will only work for this particular body of a patient and would only attack the cancer molecule without touching anything else on the site, the cancer cells.

So this is the way things will change very, very soon. And there are things like crispr, which is a gene editing tool before people get heart diseases or cancer, you can identify these are the things that will cause them and actually prevent them even from happening and edit the genome. And all this is possible because of continuous monitoring. To conclude, we are spending 17% of our GDP on healthcare today and it's going out of control. We truly really need to change this. But unfortunately we have an aging population, people who don't, we don't have enough clinicians and how do we solve this problem?

AI is a great, wonderful tool to do this. And it's not just a gimmick or something that's going to stay for the next six months and go away. This is here to stay. And people who use AI will actually outdo people who don't use AI and will generate transformative outcomes. I am the living proof for this. I started a company 12 years back. Today we take care of over a million seniors. And in the last two years, we prevented over 100,000 falls and 16,000 falls with major injuries just in the United States. I am Deepak Gadipati. Thank you.

Artificial Intelligence, Healthcare, Aging Population, Technology, Innovation, Ethics, Tedx Talks