Case study

Personalising heart failure treatment

October 7, 2019

Overview

Heart failure is a complex disease, and its character is diverse in its causes, effects and how patients respond to treatment.

Purpose

Despite this complexity and diversity, patients are treated identically with ACE inhibitors that help to improve blood supply to the heart muscle and lower blood pressure and beta-blockers that manage heart rhythm and blood pressure; with the aim of improving heart function. Within heart failure there are currently recognised differences in a patient’s heart efficiency.

A measurement system to assess this efficiency difference – known as “ejection fraction”, measures the percentage of blood leaving and entering the heart. Approximately half of people with heart failure have preserved ejection fraction, while the other half have reduced ejection fraction. For patients with reduced ejection fraction heart failure there is currently no treatment.

The work

Sensyne Health is using machine-learning-based clustering algorithms that divide ethically sourced anonymised data into smaller groups of patients with similar traits (clusters) to characterise specific trends and patterns within the subdivided heart failure patient populations (subpopulations). A newly developed deep embedded clustering (DEC) approach uses blood pressure signatures from over 65,000 admitted patients to identify and categorise different clusters of patients based on their blood pressure profile. This cluster data is then applied to heart disease to identify subpopulations to drive better patient stratification, improve clinical trial design and ultimately ensure the right patients are receiving the right heart failure drugs.

Outlook

Sensyne Health has applied this advanced data-driven approach using the single variable of blood pressure and now intends to expand the variables beyond blood pressure to a multi-variable approach using blood tests, vital signs, electronic health record data and medical imaging. This machine learning clustering work will enable Sensyne Health to identify specific “patterns” across broader, deeper and more personalised sets of data to identify specific groups of patients suffering from different forms of heart failure, and help to develop personalised treatments that are effective for particular patient groups.

Case study

Personalising heart failure treatment

October 7, 2019

Overview

Heart failure is a complex disease, and its character is diverse in its causes, effects and how patients respond to treatment.

Purpose

Despite this complexity and diversity, patients are treated identically with ACE inhibitors that help to improve blood supply to the heart muscle and lower blood pressure and beta-blockers that manage heart rhythm and blood pressure; with the aim of improving heart function. Within heart failure there are currently recognised differences in a patient’s heart efficiency.

A measurement system to assess this efficiency difference – known as “ejection fraction”, measures the percentage of blood leaving and entering the heart. Approximately half of people with heart failure have preserved ejection fraction, while the other half have reduced ejection fraction. For patients with reduced ejection fraction heart failure there is currently no treatment.

The work

Sensyne Health is using machine-learning-based clustering algorithms that divide ethically sourced anonymised data into smaller groups of patients with similar traits (clusters) to characterise specific trends and patterns within the subdivided heart failure patient populations (subpopulations). A newly developed deep embedded clustering (DEC) approach uses blood pressure signatures from over 65,000 admitted patients to identify and categorise different clusters of patients based on their blood pressure profile. This cluster data is then applied to heart disease to identify subpopulations to drive better patient stratification, improve clinical trial design and ultimately ensure the right patients are receiving the right heart failure drugs.

Outlook

Sensyne Health has applied this advanced data-driven approach using the single variable of blood pressure and now intends to expand the variables beyond blood pressure to a multi-variable approach using blood tests, vital signs, electronic health record data and medical imaging. This machine learning clustering work will enable Sensyne Health to identify specific “patterns” across broader, deeper and more personalised sets of data to identify specific groups of patients suffering from different forms of heart failure, and help to develop personalised treatments that are effective for particular patient groups.

Case study

Personalising heart failure treatment

Personalising heart failure treatment

October 7, 2019

Overview

Heart failure is a complex disease, and its character is diverse in its causes, effects and how patients respond to treatment.

Purpose

Despite this complexity and diversity, patients are treated identically with ACE inhibitors that help to improve blood supply to the heart muscle and lower blood pressure and beta-blockers that manage heart rhythm and blood pressure; with the aim of improving heart function. Within heart failure there are currently recognised differences in a patient’s heart efficiency.

A measurement system to assess this efficiency difference – known as “ejection fraction”, measures the percentage of blood leaving and entering the heart. Approximately half of people with heart failure have preserved ejection fraction, while the other half have reduced ejection fraction. For patients with reduced ejection fraction heart failure there is currently no treatment.

The work

Sensyne Health is using machine-learning-based clustering algorithms that divide ethically sourced anonymised data into smaller groups of patients with similar traits (clusters) to characterise specific trends and patterns within the subdivided heart failure patient populations (subpopulations). A newly developed deep embedded clustering (DEC) approach uses blood pressure signatures from over 65,000 admitted patients to identify and categorise different clusters of patients based on their blood pressure profile. This cluster data is then applied to heart disease to identify subpopulations to drive better patient stratification, improve clinical trial design and ultimately ensure the right patients are receiving the right heart failure drugs.

Outlook

Sensyne Health has applied this advanced data-driven approach using the single variable of blood pressure and now intends to expand the variables beyond blood pressure to a multi-variable approach using blood tests, vital signs, electronic health record data and medical imaging. This machine learning clustering work will enable Sensyne Health to identify specific “patterns” across broader, deeper and more personalised sets of data to identify specific groups of patients suffering from different forms of heart failure, and help to develop personalised treatments that are effective for particular patient groups.

Case study

Personalising heart failure treatment

Personalising heart failure treatment

Overview

Heart failure is a complex disease, and its character is diverse in its causes, effects and how patients respond to treatment.

Purpose

Despite this complexity and diversity, patients are treated identically with ACE inhibitors that help to improve blood supply to the heart muscle and lower blood pressure and beta-blockers that manage heart rhythm and blood pressure; with the aim of improving heart function. Within heart failure there are currently recognised differences in a patient’s heart efficiency.

A measurement system to assess this efficiency difference – known as “ejection fraction”, measures the percentage of blood leaving and entering the heart. Approximately half of people with heart failure have preserved ejection fraction, while the other half have reduced ejection fraction. For patients with reduced ejection fraction heart failure there is currently no treatment.

The work

Sensyne Health is using machine-learning-based clustering algorithms that divide ethically sourced anonymised data into smaller groups of patients with similar traits (clusters) to characterise specific trends and patterns within the subdivided heart failure patient populations (subpopulations). A newly developed deep embedded clustering (DEC) approach uses blood pressure signatures from over 65,000 admitted patients to identify and categorise different clusters of patients based on their blood pressure profile. This cluster data is then applied to heart disease to identify subpopulations to drive better patient stratification, improve clinical trial design and ultimately ensure the right patients are receiving the right heart failure drugs.

Outlook

Sensyne Health has applied this advanced data-driven approach using the single variable of blood pressure and now intends to expand the variables beyond blood pressure to a multi-variable approach using blood tests, vital signs, electronic health record data and medical imaging. This machine learning clustering work will enable Sensyne Health to identify specific “patterns” across broader, deeper and more personalised sets of data to identify specific groups of patients suffering from different forms of heart failure, and help to develop personalised treatments that are effective for particular patient groups.

Arrange to meet us
Case study

Personalising heart failure treatment

October 7, 2019

Overview

Heart failure is a complex disease, and its character is diverse in its causes, effects and how patients respond to treatment.

Purpose

Despite this complexity and diversity, patients are treated identically with ACE inhibitors that help to improve blood supply to the heart muscle and lower blood pressure and beta-blockers that manage heart rhythm and blood pressure; with the aim of improving heart function. Within heart failure there are currently recognised differences in a patient’s heart efficiency.

A measurement system to assess this efficiency difference – known as “ejection fraction”, measures the percentage of blood leaving and entering the heart. Approximately half of people with heart failure have preserved ejection fraction, while the other half have reduced ejection fraction. For patients with reduced ejection fraction heart failure there is currently no treatment.

The work

Sensyne Health is using machine-learning-based clustering algorithms that divide ethically sourced anonymised data into smaller groups of patients with similar traits (clusters) to characterise specific trends and patterns within the subdivided heart failure patient populations (subpopulations). A newly developed deep embedded clustering (DEC) approach uses blood pressure signatures from over 65,000 admitted patients to identify and categorise different clusters of patients based on their blood pressure profile. This cluster data is then applied to heart disease to identify subpopulations to drive better patient stratification, improve clinical trial design and ultimately ensure the right patients are receiving the right heart failure drugs.

Outlook

Sensyne Health has applied this advanced data-driven approach using the single variable of blood pressure and now intends to expand the variables beyond blood pressure to a multi-variable approach using blood tests, vital signs, electronic health record data and medical imaging. This machine learning clustering work will enable Sensyne Health to identify specific “patterns” across broader, deeper and more personalised sets of data to identify specific groups of patients suffering from different forms of heart failure, and help to develop personalised treatments that are effective for particular patient groups.