Putting Out the Fire: How Machine Learning Can Help to Mitigate Staffing Shortages That Risk Patients’ Lives

by
Marjorie Tashman
March 8, 2022
Share this post

COVID-19 has changed so many things in the U.S. healthcare system. Among them is the exacerbation of a staffing shortage that was already impeding the provision of quality care to patients. Providers are stretched thin, as they struggle to cover the parts of their practice that are shorthanded — namely, all of them — and actually treat patients. Add to this the fact that more than two-thirds of Americans have skipped regular appointments in the past couple of years due to COVID-related fears, and you have a massive population of people in need who are not receiving necessary care.


As staffing limitations reach crisis levels that have been compounded by COVID, properly triaging patients in order to provide preventive care is increasingly important — and increasingly more challenging. 


“With limited care-management resources, healthcare needs to be directed at precisely the right patients at the right moment in time,” says Teddy Cha, co-founder and CEO of pulseData. “It’s like you have a fire in your house.. You can manage with just two firefighters — if you know exactly which room to go to. But if you don’t have that information, you need to hire 20 firefighters to go into every room to give you the same chance of putting the fire out. Even better if you knew an hour beforehand to turn the stove off.”  


Now, finally, this information exists in a usable and attainable format. Healthcare has always produced an abundance of patient data. However, when it’s not been actionable, the problem persists. Too many barriers - data quality, timelines, and accessibility - make patient data hard to use. Machine-learning healthcare insights can provide data  that is consumable, accurate, and timely enough to give providers and care managers what they need for appropriate or even proactive interventions. By putting the power of machine learning in the hands of busy healthcare providers at scale, pulseData delivers what providers need to be identifying at-risk patients earlier. This means providers attain more time they need to become more successful.  Understanding through machine learning brings about new skills and capabilities necessary to pair those patients with the right care so intervention occurs in time to prevent costly decline.


“Tens of thousands of patients have died or been injured* year after year because readily available information was not used — and is not being used today — to guide their care,” Michael Millenson wrote in his book, Demanding Medical Excellence. “[The health-care delivery system] must be restructured according to evidence-based medical practice, regular assessment of the quality of care and accountability.”

*Injured meaning their conditions worsened given they went unseen.


Spotting Imminent Danger Earlier


A key part of this restructuring is a shift from reactive to proactive health care. Take chronic kidney disease (CKD) and end-stage kidney disease (ESKD) . CKD and ESKD are chronic health conditions that often go undetected for years before emerging as an urgent need for costly kidney-replacement therapy. 


Left undetected and untreated, chronic kidney, diabetes, and end-stage kidney disease, among other chronic diseases, increase costs and care demand exponentially.  According to the Centers for Medicaid and Medicare Services (CMS), approximately 20% of Medicare dollars — $114 billion a year — are spent on Americans with kidney disease .


According to the Centers for Disease Control and Prevention (CDC) more than 37 million people in the U.S. are estimated to have CKD, and another 20–25 million are at risk of developing it. What’s more, some 90% of adults with CKD do not even know they have it. Each year, 120,000 U.S. patients are diagnosed with ESKD and require either maintenance dialysis or a kidney transplant to stay alive. 


For many patients, it never had to reach that point.


Current clinical workflows are not designed to identify or monitor these high-risk patients. They rely on point-of-care management rather than proactive care, a trend that has been exacerbated by the staffing shortage. This means patients who experience a drastic decline in kidney function are often left “unseen” by a clinical team until their next scheduled visit or a hospitalization event. By then, it may be too late.


Health plans, self-funded employers, value-based care and other risk-bearing organizations need a real-time understanding of those in their population who are at-risk for these conditions in order to minimize both disease progression and avoidable unnecessary expenditures.


A U.S. patent was awarded to pulseData for machine-learning technology that uses commonly available data to predict the risk of renal decline. The technology harmonizes data sources already generated by health systems — electronic health records (EHR), claims records, ADT feeds, pharmacy data, clinical laboratory results, and complementary data sources — to diagnose, risk stratify, and predict adverse events for patients with chronic disease or disorders.  pulseData goes even further by tailoring risk scores to the capabilities and needs of each clinical team. This empowers them to prioritize the riskiest patients and recommend appropriate interventions — before it’s too late. 


To create such predictive profiles manually could take professional staff months or even years. At a time when staffing levels are already painfully low, and given the lack of access to accurate information, many patients literally don’t have the time to wait. 


By putting the power of machine learning in the hands of busy healthcare providers at scale, we can predictably pair the right patients with the right care intervention at the right time to prevent costly declines and improve lives.


Marjorie Tashman

Putting Out the Fire: How Machine Learning Can Help to Mitigate Staffing Shortages That Risk Patients’ Lives

Marjorie Tashman
March 8, 2022

COVID-19 has changed so many things in the U.S. healthcare system. Among them is the exacerbation of a staffing shortage that was already impeding the provision of quality care to patients. Providers are stretched thin, as they struggle to cover the parts of their practice that are shorthanded — namely, all of them — and actually treat patients. Add to this the fact that more than two-thirds of Americans have skipped regular appointments in the past couple of years due to COVID-related fears, and you have a massive population of people in need who are not receiving necessary care.


As staffing limitations reach crisis levels that have been compounded by COVID, properly triaging patients in order to provide preventive care is increasingly important — and increasingly more challenging. 


“With limited care-management resources, healthcare needs to be directed at precisely the right patients at the right moment in time,” says Teddy Cha, co-founder and CEO of pulseData. “It’s like you have a fire in your house.. You can manage with just two firefighters — if you know exactly which room to go to. But if you don’t have that information, you need to hire 20 firefighters to go into every room to give you the same chance of putting the fire out. Even better if you knew an hour beforehand to turn the stove off.”  


Now, finally, this information exists in a usable and attainable format. Healthcare has always produced an abundance of patient data. However, when it’s not been actionable, the problem persists. Too many barriers - data quality, timelines, and accessibility - make patient data hard to use. Machine-learning healthcare insights can provide data  that is consumable, accurate, and timely enough to give providers and care managers what they need for appropriate or even proactive interventions. By putting the power of machine learning in the hands of busy healthcare providers at scale, pulseData delivers what providers need to be identifying at-risk patients earlier. This means providers attain more time they need to become more successful.  Understanding through machine learning brings about new skills and capabilities necessary to pair those patients with the right care so intervention occurs in time to prevent costly decline.


“Tens of thousands of patients have died or been injured* year after year because readily available information was not used — and is not being used today — to guide their care,” Michael Millenson wrote in his book, Demanding Medical Excellence. “[The health-care delivery system] must be restructured according to evidence-based medical practice, regular assessment of the quality of care and accountability.”

*Injured meaning their conditions worsened given they went unseen.


Spotting Imminent Danger Earlier


A key part of this restructuring is a shift from reactive to proactive health care. Take chronic kidney disease (CKD) and end-stage kidney disease (ESKD) . CKD and ESKD are chronic health conditions that often go undetected for years before emerging as an urgent need for costly kidney-replacement therapy. 


Left undetected and untreated, chronic kidney, diabetes, and end-stage kidney disease, among other chronic diseases, increase costs and care demand exponentially.  According to the Centers for Medicaid and Medicare Services (CMS), approximately 20% of Medicare dollars — $114 billion a year — are spent on Americans with kidney disease .


According to the Centers for Disease Control and Prevention (CDC) more than 37 million people in the U.S. are estimated to have CKD, and another 20–25 million are at risk of developing it. What’s more, some 90% of adults with CKD do not even know they have it. Each year, 120,000 U.S. patients are diagnosed with ESKD and require either maintenance dialysis or a kidney transplant to stay alive. 


For many patients, it never had to reach that point.


Current clinical workflows are not designed to identify or monitor these high-risk patients. They rely on point-of-care management rather than proactive care, a trend that has been exacerbated by the staffing shortage. This means patients who experience a drastic decline in kidney function are often left “unseen” by a clinical team until their next scheduled visit or a hospitalization event. By then, it may be too late.


Health plans, self-funded employers, value-based care and other risk-bearing organizations need a real-time understanding of those in their population who are at-risk for these conditions in order to minimize both disease progression and avoidable unnecessary expenditures.


A U.S. patent was awarded to pulseData for machine-learning technology that uses commonly available data to predict the risk of renal decline. The technology harmonizes data sources already generated by health systems — electronic health records (EHR), claims records, ADT feeds, pharmacy data, clinical laboratory results, and complementary data sources — to diagnose, risk stratify, and predict adverse events for patients with chronic disease or disorders.  pulseData goes even further by tailoring risk scores to the capabilities and needs of each clinical team. This empowers them to prioritize the riskiest patients and recommend appropriate interventions — before it’s too late. 


To create such predictive profiles manually could take professional staff months or even years. At a time when staffing levels are already painfully low, and given the lack of access to accurate information, many patients literally don’t have the time to wait. 


By putting the power of machine learning in the hands of busy healthcare providers at scale, we can predictably pair the right patients with the right care intervention at the right time to prevent costly declines and improve lives.