KenSci
KenSci Menu KenSci
KenSci Close KenSci

Clinical analytics

Predict risk to manage care outcomes better

Clinical analytics
A proactive method of identifying and reducing patient risk
Delivering ROI in 12 weeks

A proactive method of identifying and reducing patient risk

KenSci’s risk prediction platform for healthcare is engineered to ingest, transform and integrate disparate sources of healthcare data, including EHR, Claims, Admin/Finance, streaming and other sources. KenSci uses machine learning to recognize patterns in large volumes of data, helping health systems view the granular details of the patient’s history and predict future risks for optimal care.

Use cases we solve for

Care variation Prediction
Chronic Kidney Disease Prediction
End of Life Prediction
Risk of Readmission Prediction
Sepsis Prediction
Care variation Prediction
Care variation Prediction
Chronic Kidney Disease Prediction
End of Life Prediction
Risk of Readmission Prediction
Sepsis Prediction
Care variation Prediction

Care variation Prediction

Variation in care is a major factor influencing increased population healthcare costs. While some variation in care is expected and appropriate for certain cohorts, variation can be a sign of decreased access for patients, inappropriate treatment, poor access to treatment alternatives, and reimbursements based ...

+ read more

Variation in care is a major factor influencing increased population healthcare costs. While some variation in care is expected and appropriate for certain cohorts, variation can be a sign of decreased access for patients, inappropriate treatment, poor access to treatment alternatives, and reimbursements based on volume and increased treatment costs.

 

The deficit

Executives and physicians face the menacing task of controlling care variation to control spending in healthcare organizations. Due to the multiple complicating factors contributing to this cost and quality driver, sophisticated prediction to proactively manage costs related to unstandardized care is an immediate need.

Integrates data sources

Integrates data sources

Integrates disparate data sources and helps users understand and control care variation using predictive modeling, and applies algorithms via the KenSci Machine Learning Platform to fully leverage the rich and underlying patterns embedded within the claims data to derive accurate and actionable insights.

Variance insight alerts

Variance insight alerts

Can alert an Executive and physician that an increased population cost is attributed directly to care variance from the use of certain kinds of supplies, providing insights that allow for opportunities to consider standardizing supply chain and logistics for products moving forward.

Continuous learning

Continuous learning

Learns from a pool of historical clinical data to drill down to determine Month over Month variation in cohort characteristics and treatment for diseases and procedures.

Maximize efficiency

Maximize efficiency

Reduces clinical variance and care utilization critical to maximizing the efficiency and efficacy of the healthcare system – ultimately bringing down costs and making care more affordable and accessible for everyone.

Chronic Kidney Disease Prediction

Chronic Kidney Disease Prediction

An estimated 23 million people in the United States, nearly 12% of the adult population, have CKD and are at increased risk for progression to kidney failure, as well as cardiovascular-related events. Clinical decision-making for CKD is challenging due to the heterogeneity of kidney diseases, variability in ...

+ read more

An estimated 23 million people in the United States, nearly 12% of the adult population, have CKD and are at increased risk for progression to kidney failure, as well as cardiovascular-related events. Clinical decision-making for CKD is challenging due to the heterogeneity of kidney diseases, variability in rates of disease progression, and the competing risk of cardiovascular mortality.

 

The deficit

Although there are proven therapies to improve outcomes in patients with progressive kidney disease, these therapies may also cause harm and add greatly to the cost of care. Currently, there are no widely accepted predictive instruments for CKD progression, and physicians must use expert judgment and heuristics to decide which patients to treat, resulting in risking treatment delays for those who ultimately develop kidney failure or unnecessary treatment in those whose CKD condition does not progress.

Customized applications and APIs

Customized applications and APIs

By ingesting and integrating the vast numbers of data points available in claims, EHR and other data sources, machine learning algorithms can detect the patterns indicating potential for disease progression far better than unaided human practitioners. Risk predictions are visualized through user-customized applications or via API’s inputted into the current end-user workflow.

Care of patients with CKD

Care of patients with CKD

Progression of chronic kidney disease is difficult to predict due to patient heterogeneity and the multitude of factors influencing progression. KenSci’s machine learning models for CKD disease progression can take in many more factors, resulting in greater accuracy of prediction and increased cost savings, even for patients presenting with milder disease symptoms.

Unified system of records

Unified system of records

KenSci’s platform provides a single system of record, integrating all available data sources into a common schema that can then be applied to KenSci’s Machine Learning models for CKD Progression prediction.

Trained for accuracy

Trained for accuracy

KenSci’s ML models have been trained for accuracy on millions of data elements. By accurately predicting declines in renal function, hospitals, physicians, and patients can better manage the care of patients with CKD.

End of Life Prediction

End of Life Prediction

Few domains in healthcare offer a clearer opportunity for realization of the Quadruple Aim than care at the end of life. This period is all too commonly marked by aggressive, invasive intervention representing a significant burden to patients, their loved ones, and the broader system of care – yet without...

+ read more

Few domains in healthcare offer a clearer opportunity for realization of the Quadruple Aim than care at the end of life. This period is all too commonly marked by aggressive, invasive intervention representing a significant burden to patients, their loved ones, and the broader system of care – yet without commensurate improvement in outcomes or quality of life.

 

The deficit

Fundamentally, the root cause of this status quo is a deficit in prediction. While a physician may be aware of a patient’s worsening, the variable course keeps the clinical team and the patient focused on the hope of improvement, and hesitant to talk about mortality prediction. But hospice referrals are the rare situation in medicine where the physician needs to “predict mortality within 6 months.” Not only is this prediction difficult one patient at a time, and clinicians have been shown to have low accuracy in doing so, patients and families may not be interested or ready for the conversation.

Predict with large data sets

Predict with large data sets

The KenSci solution identifies patients at risk for mortality in 6-12 months with 74% accuracy using a weighted classifier to strengthen predictability among large sets of data.This machine learning approach increases the reliability to answer the question of who might die with the next year and help clinicians make careful decisions about the most appropriate care based on patient and family wishes.

Unified system of records

Unified system of records

The KenSci Platform ingests all disparate sources of data into a single System of Record where KenSci’s ML models utilize claims and EHR data, including demographics, diagnosis and utilization history, labs and vital signs to provide insights only available by leveraging Machine Learning.

Singular insight dashboard

Singular insight dashboard

An interactive dashboard indicates patient risk scores and trajectory over time, alongside factors contributing to risk, serving as an at-a-glance clinical tool for ease of identification for patients at risk for mortality.

Trained for accuracy

Trained for accuracy

KenSci utilizes multiple ML predictive models to determine the probability of a patient dying within 6 - 12 months from the date of prediction to optimize end of life care. KenSci’s ML models have been trained for accuracy on millions of data elements.

At risk stratification

At risk stratification

This solution provides risk scores to individual physicians and care managers to identify at-risk patients to enable appropriate care prioritization, and the risk predictions are visualized through user-customized applications or via API’s inputted into the current end-user workflow.

Risk of Readmission Prediction

Risk of Readmission Prediction

Despite recent improvements, unplanned hospital readmissions continue to be a massive public health problem in the United States. Readmissions impact up to 3.3 million patients in the US and cost an estimated $26 billion per year, with an estimated $17 billion being potentially avoidable. The Center for ...

+ read more

Despite recent improvements, unplanned hospital readmissions continue to be a massive public health problem in the United States. Readmissions impact up to 3.3 million patients in the US and cost an estimated $26 billion per year, with an estimated $17 billion being potentially avoidable. The Center for Medicare & Medicaid Services (CMS) now publicly reports rates of readmissions, and financial penalties are assessed to hospitals with high readmission rates.

 

The deficit

The causes for failure to prevent most readmissions are complex. Patient demographics, diet, and medical co- morbidities exacerbate this profound problem. Hospital treatment, medications, and discharge and follow-up processes also appear to have an impact. The actual interaction amongst this vast array of variables is not yet fully understood. Unfortunately, guidelines recommend population-based interventions with little to no consideration of individual patients, and clinicians are provided insufficient knowledge about clinical decision support.

Answer harder questions

Answer harder questions

The KenSci solution identifies patients at risk for a preventable readmission with 71% accuracy using a weighted classifier to strengthen predictability among large sets of inpatient data real- time. This machine learning approach increases the reliability to answer the question of who might return to the hospital within thirty days of discharge.

Identify at-risk patients

Identify at-risk patients

KenSci’s has developed a predictive model to identify currently hospitalized patients at the highest risk of all-cause rehospitalization in the next 30-days. This solution is utilized by individual physicians and care managers at the bedside and provides risk scores for identifying at-risk patients for prioritization of appropriate multidisciplinary care coordination.

Learn from existing data

Learn from existing data

Variables used in the prediction are an aggregate of labs, comorbidity history, demographics, and patient inpatient history from the electronic medical record during the hospital stay.

Singular insight dashboard

Singular insight dashboard

An interactive dashboard provides patient risk scores serving as an at-a-glance clinical tool for ease of identification for patients at- risk for readmission.

Sepsis Prediction

Sepsis Prediction

Recognizing the need to predict those most at risk to develop sepsis and reduce the costs associated with their care is vital. Sepsis accounts for 750,000 hospitalizations annually. Annual costs of care for Sepsis in the US equal approximately $20 billion annually, more than any other single disease. The annual...

+ read more

Recognizing the need to predict those most at risk to develop sepsis and reduce the costs associated with their care is vital. Sepsis accounts for 750,000 hospitalizations annually. Annual costs of care for Sepsis in the US equal approximately $20 billion annually, more than any other single disease. The annual mortality rate due to sepsis is between 15% and 30% – with as many as 200,000 deaths in the US alone each year.

 

The deficit

Sepsis is the leading cause of hospital readmission among the adult population in the US. Severe sepsis and its progression to septic shock have been difficult to predict and treat promptly. Beginning in 2017, the Centers for Medicare & Medicaid services (CMS) began adjusting hospital payments based on the quality of sepsis care. The KenSci solution for Sepsis Prediction allows hospitals and care givers to predict those most likely to develop insepsis care

Learn more from data

Learn more from data

KenSci’s Platform and ML Sepsis Risk Prediction Solution uncovers insights that lead to information regarding which patients are at highest risk to get patients the care they need sooner.

Unified system of records

Unified system of records

KenSci’s platform provides a single system of record, integrating all available data sources into a common schema that can then be applied to KenSci’s Machine Learning models for sepsis risk prediction.

Trained for accuracy

Trained for accuracy

KenSci’s ML models have been trained for accuracy on millions of data elements. They factor hundreds of variables simultaneously, resulting in greater accuracy of prediction, at an earlier time point, with increased cost savings.

Easy insight visualization

Easy insight visualization

Risk predictions are visualized through user-customized applications or via API’s inputted into the current end-user workflow.

Stay one step ahead

  • Predict risk & reduce clinical variance

    Predict risk & reduce clinical variance

    Physicians and Care managers who are dedicated to scale their mission to save lives and are frustrated by state of healthcare.

  • Improve outcomes and quality measures

    Improve outcomes and quality measures

    Machine Learning experts and computational researchers driven by a burning desire to tease life-saving insights out of data.

  • Better discharge planning & load prediction

    Better discharge planning & load prediction

    ML Engineers who have shipped Cloud OS and Search Engines. Now creating the most meaningful platform of their career.

Get in touch