Save Roomer Name with airSlate SignNow

Save roomer name

we can go ahead and get started my notes the last session of the day so I Lisa she's everyone taking an interest in data mining at 4pm my name is Anne lane' I'm with Public Health Management Corporation PMC is a human services public health institute located in the Philadelphia region we provide human and health services to about 125,000 clients in the region I'm here in lieu of my boss Mike Ford ocean and I'm sorry that he couldn't make it cuz he is one of those people that has a genuine passion for this and he is someone who actually makes you know health informatics really exciting which is rare but we're going to talk about actually I'm going to let me introduce himself at night for tech first great so first a lot first off thank you everyone for being here my name is z and i'm the CEO by karthik were based in cambridge massachusetts and essentially we are a company developing health IT solutions specifically data mining for the poor analytic solutions to help provider systems deliver better care and help control costs as well we work with health insurances we work with providers both acute facilities as well as hep-th seas and we also work with state governments so we're going to talk about a model that may be able to reduce costs and improve outcomes at the same time it it's data mining to predict future super utilizers of expensive health services so step one is is defining what a super utilizar is different than a super user like of an EHR which is an administrator a super utilizar is usually a patient who over tends to over use health services typically the most expensive health services in the most expensive location particularly the emergency department they often have comorbidity so they're dealing with multiple chronic illnesses so they don't just have diabetes they have diabetes and hypertension and heart disease they have a need for more than just medical services they usually need social services as well for example a patient with diabetes might need a pair of eyeglasses because they can't read the small numbers on their their glucose meter which is causing them to mismanage their blood sugar and they can't afford eyeglasses and someone being around to know that they could use I glasses as evidence of the lack of coordination that is that feeds into super utilizers they could have a need for mental health and addiction services and nutrition nutrition services since some of the most prevalent chronic diseases are often negatively impacted by poor diet there's a disparity in the use all clients are not the same so the solution that we treat most patients with is not as effective for a super super utilizar the Robert Wood Johnson Foundation just released two million dollars in grants to address super utilizers and the lack of coordination that seems to yeah that that tends to to cause them so the way that super utilizers are being addressed right now is that we load the neediest clients with intensive resources and we all know that this is an extremely expensive approach to health care the the 8020 rule is certainly applicable here except with super utilizers it's more it's closer to about 5% of the sickest patients that can cause upwards of sixty percent of the cost of healthcare services and spending some medical organizations have become so used to the churn and burn of the fee-for-service model when it comes to super utilizers that they don't know how to break the cycle because they're so used to their that revenue and moving to a paper crew a pay-for-performance model can be difficult so in talking about a new solution the question is can it be done differently and the answer seems to me that it has to be done differently with the sicker minority distorting the usage model this it's just not a sustainable especially for community health centers so there's a four-part solution that we're proposing part one is data analysis and using data mining to look at profiles of people that we know are already super utilizers if you work with a community health center you know that you can ask your clinicians who the frequent flyers are the people that they're seeing on a regular basis and then look at you know some of their demographics and their informatics to see if you can predict some of the future super utilizers that might in the future be over utilizing services in this regard health centers are a great catalog of patients with comorbidities and we know that we can't do any of this without a bank of data that means that a current a health center would have to have an active EHR and they would have to be actively monitoring the data that goes into it as far as quality the garbage in garbage out motto motto here certainly would apply fortunately programs like the meaningful use incentive program would be helpful here particularly because they're ongoing stage one of meaningful use is coming to a close but that doesn't mean that the the quality program is over we have sage two coming up so we know that there is an emphasis on quality for our EHRs which means that the data is relatively reliable and could help us identify super utilizers in the future as well right so really part one is so part one the entire tent and tender part one is take what clinicians already know about super utilizar behavior and what's contained within the existing data as recorded in that electronic health record systems and to extrapolate about key characteristics of finding super utilizers such that you might be able to predict future supervisors before they are officially categorize as such and that's what our team specializes is and specialize in and focus on or most of our team consists in stettin or sand mathematicians and what we've discovered is the best way to identify super utilizar is actually through mining clinician notes and the predictability from mining just sentiment of what we call sentiment analysis by mining with the physicians where clinicians say about the patient behavior and the patient's mood etc are actually very predictive of their super utilizar status and this is in contrast with the existing approach and essentially how we embarked upon this is was through a joint initiative with ph and see but we got prompted by some of the larger health insurances in in greater philadelphia that said you know hey super utilizers are really a financial drain for everyone involved and so is there a partnership or is there an opportunity for us one to proactively identify them and then to to do something about it and this is our proposed model is in contrast to wear a lot of the current efforts are very much retrospective there are people already using our rooms 40 to 50 times a year on average and you know they get recorded the health plans have their information send it to the agency and then ask them to do something about it the second part is using the cure a care management feature we envision this probably to be as a nurse coordinator an RN care coordinator with an appropriate case load of this cohort of predicted super future super utilizers would target extra preventive resources and the idea is that the fixed cost of the nurse's salary could be sustained by the preventative resources and this isn't to say that you would have to hire a new manager this is just sort of redirecting resources that are already sustained within the clinic and the important thing to keep in mind here is that medical services can't treat the origin of the problem and then it's almost too late by the time that you're having to intervene medically so this is a pre-intervention with preventive services that's all managed by a nurse coordinator within within a clinic to the targeted population and then part three is really extending the services outside of the walls of the clinic so it says services in the home but it really even goes beyond the home so obviously one part of that would be a home monitoring health monitoring from home this graphic right here is from a company called manyata and they actually would utilize your cable TV box for health monitoring so it would be you know pretty user friendly and patient friendly and you can probably see there's a real connection for opportunity with public housing if you have sort of a captive audience of these patients with the comorbid diseases all within you know the same housing development there's an opportunity to coordinate care for those for those patients health education services can also be implemented pretty easily in a in a public housing model if you know there's a classroom if there's a classroom space a nurse coordinator would easily be able to provide targeted health education and then connecting those residents to other health services like nutrition management behavioral mental health services addiction services that all conveniently can be administered within a community health center partnering with a public public housing development reina so from our perspective is also that corn to extend this beyond the health clinics itself and into the home environment for several key reasons one it's really hard to get longitudinal comprehensive data about a patient even for health insurance and unfortunately if you don't have the complete data representation of a patient it's really hard to identify what's missing or make predictive proactive decisions so that if you could really extend your services such that you know you're following the patient home and into their daily routine daily activities you can be more become more effective in addition to that I think just that I be analytical part of it it's important to keep in mind it's not only used proactively to identify the profiles of super utilizers but it's also used on an ongoing basis to reply the processes as kind of feedback loop of what's happening to develop benchmarks and to quantify and the ROI of the results and then the last part that's really critical is just data data analysis so making the connection between the coordinated care for this targeted group and a reduction in you know unnecessarily visits the emergency department this is making the case for the cost-effectiveness of using a nurse coordinator and then thinking about what the measures of success would be was there you know the reduction in the emergency department over usage was the cost of care per patient lower with this model than if no intervention have been implemented at all and then you know finally most importantly is the patient healthier right and this roi component is very important because there are health insurances out there health plans out there that's willing to support this model we've been talking to several along with HMC and the key point to getting them involved either to subsidize we're in centralized creation of this program is they really need to see the quantified roi we need to know that this is effective and this is somehow helping to manage the super utilizar population at the end of a the super utilizers are actually huge green a lot of these help plants especially after the upcoming ACA Thanks yep and so using the data analytics the show that the program is working as a vishing successful benchmarks that's a very important part of the process so why would why would a health center implement something like this the proof would be obviously in reducing costs and seeing an improvement in outcomes and the goal would be enhanced reimbursements and payments from pairs but it's important to note that the goal is both reduced costs and improve outcomes and pears would obviously take an interest in this model as you said with the upcoming implement about Care Act and will implementation have medicaid with you know the Medicaid HMOs looking at a broader covered covering a broader base of the sickest population it's a it's a it's a positive model to implement and then the great thing about this model is that once you know it's been tested and proven is that it can be replicated throughout the country so this is just sort of an overview of the the program process so use using the existing data that health centers have been collecting through their EHR with an attention to data quality and input using the clinical intervention with a nursing care coordinator and then deploying the analytics to support the program evaluation quantifying the ROI of the nur of a nurse coordinator salary and then hopefully obtaining your support and scaling the model being able to replicate the model throughout the country so yeah my final graphic here is mostly just bringing it back to the quality loop so the goal is always you know from a clinical perspective is better outcomes healthier patient you can't you have to be able to sustain your clinic number one and you have to be able to lower your costs just sustain your clinic and hopefully that would come with enhance these enhanced payment right and the corresponding hita component health IT component is really utilizing these clinical systems and a lot of help I helped a lot of clinics are already implementing and truly to to truly use them in the NEPA way such that you're improving quality and reducing the cost of care for a broad number of players involved and also you're really using the H IT environment to provide feedback on what's happening in the community who's you know who are your most troublesome population and what are the most effective way to intervene with them so really the focus on sustainability thoughts question we're hoping to work with ya right so actually if you are so if you go back to the first slide this is what we're doing specifically with HSE and of course it makes a lot of sense to reach out to the community both to the acute care facilities and to the health plan so that you get the comprehensive picture of what's happened with your patient but that's not realistic starting to go and the goal is to prove validate the model enough such that you can start building support within the community so you can start getting the hospital's to share the art because they're also under a lot of financial pressure from various reforms and also from CMS to reduce ER super utilization and they're not there are acute care facilities the entire model is to treat the acute episode and more or less forget about the patient they're not very good at doing that so that if they see a player in the community that that is effective at doing this one I'm doing it proactively and two very very effective with the intervention component of it and also cost-effective there are oftentimes very interested in helping to scale this model and even though we haven't implemented this just yet with pH and see we already have had conversations with acute hospitals in the community of hey Chelsea and they're interested in working with us to see if this works so your question is how would people without insurance coverage be included well with this model focusing on community health centers who are you know if they're federally qualified they're required to provide services to folks that are covered you know whether or not they're whether or not they have coverage so I I don't know if it would mean that if they would just fall outside the cohort or not right I think that's also very important that's where you when you start expanding it within the community that's very important because a lot of even if they're homeless a lot of these patients end up and hospitals and that essentially have to write off his care as charity care so that's where again you know mutually step is kind of incremental you're trying to build support with the health insurance companies you're trying to build support with the local hospitals and it's very important to to have the analytical point to prove to them but this is an effective model and again referring back to the phn see component there's a major health system that takes a lot of patience is patient and you from theirs is now they're essentially running off a lot of charity care so they're very interested in this model with the belief that if P agency successful and according to successful in this it will help reduce their financial burden right so well the immediate cost would be you know the the salary of the care coordinator if you choose to use nurse and then subsequent costs of you know I cortex services for DNA analytic right so the software or the solution that we're developing takes a data from your EHR and we we are developing a cloud platform so the prot DP process on RN is on the cloud for the healthcare organization involved it's essentially it's our solution is designed to minimize the the technical burden on the healthcare organization there needs to be minimal infrastructure and place an effective EHR system such that the data is digitized however we recognize that you know with a lot of these clinics the infrastructures my extremely robust and even the data in there is sometimes data hiding as a problem so we're writing the the algorithm and the models to be able to handle that and again the concept goes to scaling this one step at a time so really the go at this time is to prove that there is this is a more effective program than what is currently implemented so specifically I can say what's happening with agency we actually pull the data from hmc so we develop a pipeline into a chance to use database and it's a it's a fully automated system so we get a data refresh every right now it's set up once every week and so basically there was a much hita engineering on HSN yeah I mean it's going to be typical program program implementation costs you're going to have you know part of your administration overseeing it and you know working with I Coretech to review outcomes and you have the care coordinator salary you have the costs of partnering with any other social services right and what one of the ideas is that you know there is motivation for a payer to invest in this so really ideally the a payer would be covering the cost for that right I think in terms of the discussion we've had in regards to this model really the cost comes from the kind of innovating and we're engineering the H IT solution part of this and for the clinical intervention part of it essentially what the model is more effectively using the clinician resources the clinics have by type by being able to identify the patients that clinicians seem to proactively spend time with and again like book and and Alex mentioned at this time or running this on a pilot basis and it really is to get this off the ground such that the payers and even the hospital systems will start supporting this and they've already given you know they've already heard this proposal and and they'd like to see if it works your frequent flyer right right right exactly a point and so that's why we're so excited to work with agency because from an analytics perspective really the clinicians already have a lot of the insect it's really just to translate that into a more qualified mathematics such that you know you can start going to health insurance this and start going to neighboring hospitals and save as you know hey it's not simply what we believe there's evidence for this right are you a hospital and you know it sort of goes back to that quality loop you have to make sure the data that's going into your EHR is is polished so that when you're extracting to to identify those people then you can use those metrics to predict right and we often find that based upon our own work but there's huge amount of data quality issue and actually HR systems the more expensive they are unfortunately the worst they are added there's no data Polly filtered so you know there's no automatic pump boarding the collation this is a bad entry or anything like that and there really isn't that much motivation to address the data hiking issue until you have effective clinical intervention program that's kind of based upon data mining so really you struck need to start using the data before the data will get better so metric in terms of what are the identifiers right right so a huge part of it is so we haven't officially started this program yet but on our side we Burton started playing around with vacancies data what we found is that in contrast super utilize our their clinician note in contrast to or cortical normal patients there's huge there's vastly greater number of negative sentiments expressed within those clinical notes and factors such as you know has as has his patient I've been to a clinic in the recent ex number of months and this is patient have comorbidity there's a in combination with other factors are actually also pretty good predictors of a year super utilization we were actually quite surprised about the sentimentality analysis but when not when we start looking at that it was a significant contrast and we believe that scaling this forward simply using message lethality analysis would be really effective so there's an incentive to identify them whether or not you know they have coverage because they are costing services to the health center show you be responsible for right right I think if I a necropsy or clinic is interested in discipling this patty we definitely would be pretty eager to talk to you guys and what we've been trying at this time we've been engaged with the health plans for different projects unrelated to FQHC work and so we directly went to them and said you know hey would you be interested in funding this model and and most of these pairs are in Greater Philadelphia and and they are fairly receptive so right now is kind of we've already started playing with the data so we've been creating the model and we're pretty confident that this can work and it really is now organizing things finding where the funding is for the next step and going from there you