NoviGuide

Nearly half of all children who die in the world, die in the first month of life.  Simple neonatal care protocols to address the most common causes of mortality are published in large books.  Yet too often these books remain on the shelf because the knowledge they contain is not easily applied to patient-specific cases and within site-specific resource constraints.  NoviGuide is a comprehensive clinical decision support platform that helps health systems convert text guidelines into interactive assessments.  NoviGuide's Algorithm Architect is specifically designed for medical decision trees with complex neonatal drug dosing, feeding and respiratory distress algorithms built into its Clinical Library.  Algorithms built in the Algorithm Architect are packaged as a point-of-care application; bringing expertise to the bedside while gathering data to guide system-level interventions. Scaled globally, NoviGuide would ensure that our global knowledge on the best care for all babies, reaches every baby.  

2.5 million children die in the first month of life every single year.  These deaths are not distributed evenly across the globe; babies born in poor countries with slow progress in reducing neonatal deaths account for 80% of the global burden.  Why babies die is not a mystery. Treatments for the most common causes of neonatal mortality are printed as large guideline books and are readily available online.  Many of these treatments are available and affordable, even in the poorest settings.  Why then is it so difficult for clinicians to follow guidelines?

1. Guidelines are difficult to integrate into real-world clinical care

2. Sites have varying resources for diagnosing and treating patients

3. Following the guidelines creates additional work for busy clinicians

4. Knowledge and resource barriers to following the guidelines are difficult to detect and remedy.

NoviGuide is a clinical decision support software that translates complex medical guidelines into point-of-care decision support.  The NoviGuide has four components: Algorithm Architect, Clinical Library, Point-of-Care Application and Insights Dashboard.  The Algorithm Architect is a co-design platform where health administrators convert disease-based treatment manuals into patient-centered assessments.  If the Algorithm Architect is a blank page, the Clinical Library provides the templates--allowing health administrators to build on validated algorithms populated with time-saving and error-reducing drug-dose and feeding calculators.  The pathways created in the Algorithm Architect feed directly into the Point-of-Care application on either a tablet or web app.  Holding the Point-of-Care application, clinicians can now enter in their site specific resources to receive tailored guidance for optimal care.  The Point-of-Care application is a bedside tool, helping clinicians apply knowledge derived from the guidelines to individual patients.  Each interaction with the Point-of-Care application becomes a check-up on both the patient and the heath system.  Clinicians generate and transmit data about resource availability, but also data on how those resources are applied across diverse patient scenarios.  This data is visualized in the Insights Dashboard where health administrators can detect the knowledge and resource barriers that prevent clinicians from following the guidelines.    

The NoviGuide is designed to help any baby, but our specific focus is babies born in settings where there is either a healthcare worker shortage or lack of access to expert neonatal specialists.  Our product's users are the generalist physicians, midwives and nurses who care for these babies around the world. The NoviGuide was inspired by the year and a half our medical director spent living in the eastern Democratic Republic of Congo working side-by-side with Congolese nurses and physicians to deliver neonatal care in the North Kivu capital of Goma.  That experience informed every aspect of the NoviGuide prototype including its content, drug-dosing functionalities, configurability and user interface.  The prototype then underwent human factors testing with frontline clinicians in rural South Africa and acceptability/feasibility in a district hospital in Tororo Uganda.  These research studies generated the qualitative and quantitative data needed to transform NoviGuide from a neonatal mobile application into a platform for creating and deploying clinical decision support software.  We know from research studies that NoviGuide meets the needs of frontline clinicians caring for newborns while improving their knowledge and confidence in newborn care.  Additionally, frontline clinicians report that NoviGuide is an intuitive software that reduces medical errors.

The NoviGuide directly aims to expand access to high-quality, affordable maternal and newborn care.  It addresses one of the most unyielding aspects of neonatal mortality--how to bridge the gap between what is known and what is done. The NoviGuide works on this core issue from multiple angles.  First, it converts bulky reference books into point-of-care support.  Second, it allows frontline clinicians to tailor the guidance they receive to the resources they have on hand.  Lastly, it generates data about how the guidelines are applied and pinpoints obstacles that prevent clinicians from delivering optimal care.   

NoviGuide 2.0 is decision support software designed to optimize facility-based care by healthcare practitioners in low- and middle-income countries (LMICs).  In the neonatal health domain, our main competitors are the Safe Delivery App, ThinkMD and NeoTree.  There are three features that make our solution different than our competitors: the transparency of the underlying algorithm logic, the configuration of that logic to site-specific resources and the adaptability of our platform to health areas beyond maternal-child health.

Algorithm transparency:  The NoviGuide Algorithm Architect allows health systems to see the logic underlying the software applications clinical recommendations and guidance.  This functionality dramatically reduces the risk of clinical errors in the software, while enabling health systems to proceed with confidence into more complex medical topics.  

Site-specific configuration: When clinical sites first receive the NoviGuide, they complete a resource inventory survey.  This information is used to generate a site-specific NoviGuide, one where the guidance is tailored to real-world resource constraints.  Any changes to resources, either as new equipment added or loss of previously available resources, can be added at any time.  To our knowledge, we are the only global health decision support software with this functionality.

Adaptability:  While neonatal algorithms were the first to be added to the NoviGuide Clinical Library, the Algorithm Architect allows for the co-design  with subject specialists of any medical algorithm.  As a result, the NoviGuide platform can be expanded beyond maternal-child care to match the diversity of patients seen by frontline healthcare workers.  

The NoviGuide platform is built in one of the most widely used programming languages in the world, JavaScript.  The user interface is presented using the most popular JavaScript framework, React.  We use React both for web applications (the "configurator" and "dashboard") as well as the point-of-care tablet application (which uses the React Native variant).  We developed an efficient notation format for the complex conditions needed to capture clinical logic, which we have open-sourced as "PickLogic" on Github.  On the backend, we use the Node.js Express framework and the Firebase database to store configuration and clinical data.

The NoviGuide has been evaluated through research partnerships with leading American universities and African partners.  NoviGuide 1.0 was an end-user application with three comprehensive pathways for neonatal care that underwent human factors testing in Kwazulu-Natal, South Africa.  The second iteration, NoviGuide 1.6, included an improved user interface, an expanded Clinical Library and a prototype of the Insights Dashboard.  It was tested at Tororo District Hospital in Uganda with a 13-month deployment of the software.  This research study was a collaboration between the NoviGuide team, the University of California San Francisco Preterm Birth Initiative and the Infectious Disease Research Collaboration in Uganda. The main findings from the study were (1) NoviGuide was used voluntarily by midwives who completed 1412 assessments (2) midwives reported high levels of satisfaction with the software and reported that NoviGuide prevented mistakes and (3) provider mean knowledge assessment scores improved significantly from baseline.  These results will be published by the Ugandan Principal Investigator Dr. Mary Muhindo.  Having demonstrated that NoviGuide was acceptable and feasible, it is now being deployed in a four site study in Tororo Uganda to determine if the software can assist in the uptake and usage of neonatal oxygen saturation monitoring.   

The NoviGuide is a comprehensive clinical decision support platform that helps health systems convert dense medical guidelines into intuitive, patient-centered assessments.  NoviGuide is not a simple digitization of generic guidelines.  Instead, it prompts healthcare workers to enter patient-specific clinical characteristics and delivers guidance that applies best-practice guidelines to individual cases and within site-specific resource constraints.  Equipped with the NoviGuide, healthcare workers can ensure that the care they deliver, including error prone drug dose and feeding calculations, aligns with known best practice standards and institutional guidelines.  In addition, NoviGuide's point-of-care assessments help healthcare workers proceed systematically through assessments, addressing critical issues first and ensuring that each assessment is complete and comprehensive.  With each patient assessment, healthcare workers generate valuable data about the health system--where resources are located and how those resources are applied to individual cases.  Our long-term goal is to improve the quality of patient care by helping healthcare workers apply the global knowledge of disease to individual patients while identifying where system barriers are preventing delivery of the highest standard.    

The NoviGuide was for 1412 patient assessments in a recently completed research study in eastern Uganda.  It will be used in a four site study in Tororo Uganda at facilities delivering a total of approximately 6000 births per year.  The study was slated to commence in early 2020, but was delayed because of COVID-19.  Anticipated start date is now summer 2020.  In partnership with a leading international non-governmental organization, we have submitted applications to place the NoviGuide in 15 sites in the Kirehe District of Rwanda, including the Mahama Camp.  Lastly, following emergence of the COVID pandemic, our software was featured by ICTworks as a technology to consider in the response.  We are actively pursuing partnerships to build COVID algorithms and incorporate them into our existing maternal-child health library.  

To date, NoviGuide has reached approximately 1400 beneficiaries. We anticipate that within a year our software will be used in 10,000 patient encounters.  In five years, we aim for the software to be used in 150,000 patient encounters per year.

Nearly every single policy and academic paper on neonatal mortality begins the same way: it states the size of the global neonatal mortality burden while noting that these deaths are preventable, occurring from diseases with known treatments.  This same pattern is not unique to newborns, repeating across medical specialities.  Our goal is to take aim at one of the largest quality gaps in healthcare, the gap between what is known and what is done.  Our work sees medical guidelines not as ready-to-use instructions, but as blueprints for designing point-of-care software that can be seamlessly integrated into complex patient care workflows.

We are pursuing two strategies to achieve our goal.  First, we are increasing the value of our software to health systems by growing the breadth and depth of our Clinical Library.  We believe our current library can assist healthcare workers with more than 95% of neonatal assessments, including complex logic for preterm feeding and drug dosing.  We are adding maternal care algorithms to this suite, using our experience to select clinical targets amenable to a decision support solution.  Our maternal prototype will include logic for preeclampsia, hypertension, maternal fever and gestational diabetes.  We are researching opportunities in non-communicable diseases, pediatric asthma and COVID-19.  Second, we are partnering with leading international organizations to implement NoviGuide at a scalable unit, most likely a health district.  In doing so, we aim to demonstrate NoviGuide's potential to guide system level interventions while building relationships with health ministry customers.    

The following are specific barriers that may limit our impact:

Barrier #1, Exposure:  Most digital health software used in low- and middle-income countries is created for community healthcare workers.  These softwares contain relatively simple triage assessments.  Many settings lack experience or exposure to more complex software designed for doctors, nurses and midwives.

Barrier #2, Evidence: The global evidence base for clinical decision support software is limited.  Research is largely focused on tools with narrow capabilities built into highly restrictive electronic health record environments in high-income settings.  The tools that have been evaluated are almost always within adult internal medicine.

Barrier #3, Initial funding: Those settings or countries with the highest need for decision support software--settings that lack a well-developed and specialized healthcare workforce--are often those least able to use health funds for its initial widespread deployment.

Barrier #1, Exposure:  We plan to address lack of exposure to clinical decision support software by partnering with international organizations with active relationships with health ministries and their digital health efforts.  Through these partnerships, we aim to better understand how digital health work is structured at the national level and how best to align clinical decision support software to national goals and priorities.  In addition, we aim to learn from established international partners how health ministries are measuring their progress and how to integrate NoviGuide's data into existing data collection efforts.  

Barrier #2, Evidence: We are addressing the evidence barrier by rigorously evaluating the NoviGuide with academic partners as it is designed and deployed.  We are currently partnering with the Preterm Birth Initiative at the University of California San Francisco to evaluate NoviGuide's impact on the rollout of neonatal oxygen saturation monitoring.  We also partner with the University of Connecticut's Institute for Collaboration on Health, Intervention and Policy to evaluate NoviGuide's implementation and fit for diverse settings.

Barrier #3, Initial Funding: While we believe NoviGuide can ultimately reduce costs and improve health outcomes, we recognize that health systems with limited resources will want to see the impact of a district-wide rollout before investing.  We are pursuing partnerships, philanthropic funding and grants to demonstrate NoviGuide across a health district.  

N/A

Our team includes 5 full-time staff and 3 part-time staff.  We use contractors--most commonly physicians and pharmacists--to assist in the development of the clinical algorithms.  

Our team brings together a combined 15 years of field experience and expertise in clinical medicine, technology and implementation.  

Our medical director is Joshua Bress, MD.  Joshua is a practicing pediatrician whose primary focus is the care of neonates.  He is also an Assistant Professor at the University of Connecticut's Institute for Collaboration on Health, Intervention and Policy.  He graduated from medical school at Vanderbilt University in 2007 and completed his pediatric residency at the University of California San Francisco. From 2011-12 he lived in the eastern Congo delivering pediatric and neonatal care alongside Congolese colleagues.  

Our Director of Software Design and Development is Elon Danziger.  Elon has ten years of experience shaping, building and delivering applications in digital health, education and other field.  He has led each stage of NoviGuide's development, from its beginning as a mobile application to its current form as a comprehensive clinical decision support platform.  

Our Program Manager is Jean Armas, MPH.  Jean leads our efforts in implementing and evaluating the NoviGuide.  She graduated from Emory University with a Masters of Public Health in Applied Epidemiology.  She has numerous experiences in designing and implementing technology-enabled initiatives and has been recognized for her work using digital technology to improve care for survivors of sexual violence.  

University of California San Francisco Preterm Birth Initiative & Infectious Diseases Research Collaboration:  We work with Principal Investigator Dr. Mary Muhindo to evaluate the NoviGuide in the Tororo District of Uganda.

NoviGuide adds value to health systems by making it easier for frontline healthcare workers to follow evolving clinical guidelines while seamlessly capturing data that can guide system-level resource allocation.  Our key customers are health systems that operate at a scale where care standardization becomes a critical priority.  We provide our customers with a platform for transforming dense clinical guidelines into intuitive point-of-care assessments on either a mobile device or web application.  

Our customers need our product for two reasons: there is an enormous cost when the care that healthcare workers deliver falls short of accepted standards and current methods to standardize care are ineffective and difficult to monitor. Poor care quality can cause immediate harm to patients, reduce trust in the healthcare system and demoralize healthcare teams.  Worsening illness caused by poor care quality can increase the financial burden on health systems when care is escalated and patients require medical transport.  Additionally, patients who receive poor care are more likely to return with the same problem.  Payment is increasingly linked to care quality in health systems with third party payers.  Third-party payers may refuse to pay or lower payments for care required as a result of poor care quality.  

Healthcare systems that are willing to invest in a clinical decision support platform expect that platform to address broad health issues across their target population.  We plan to build our suite of clinical pathways--the Clinical Library--through donations and grants.  To date, we have received funding that has enabled us to build a comprehensive neonatal suite and are in the prototype stage for maternal care algorithms.  We anticipate that a full suite will include heart disease, diabetes, asthma, COVID-19 and others. We plan to continue to pursue research collaborations to introduce our product to potential customers and demonstrate its impact as the full algorithm suite is being developed.  

Once we can offer a Clinical Library with clinical pathways mirroring the breadth of community health issues, we will offer our software as a subscription.  We anticipate the product will be used in low- and middle-income countries as well as rural areas of high-income countries.  We will use a tiered pricing to lower the barrier to entry.  Lower tiers will be oriented towards towards point-of-care decision support aligned to international standards, such as the World Health Organization.  Higher tiers will incorporate data analytics, white labeling and customization.  

We are applying to Solve to find partners and supporters who can help us with the following:

1. Transitioning from externally funded research studies with academic partners to offering our product directly to health systems.  
2. Thinking through various options for scaling our software, whether through licensing, bundling with other products or direct sales. 
3. Deciding the right balance between further enhancing the product and selling its current iteration.  
4. Building a customer base among large health networks and health ministries.
5. Developing the right cost structure for the product.
6. Thinking about how to work with large international non-governmental organizations.

NoviGuide was built to address a problem we encountered in our own fieldwork: it is hard for doctors and nurses to follow clinical guidelines and even harder to tell remotely if and how they are being followed.  NoviGuide was designed to solve an internal problem, not a global one.  But as we did our work, we became passionate about decision support software and its potential to address global health problems.  Now we have a product that has immediate appeal in that it addresses a near universal and thorny problem, but we do not yet have the knowledge and business model to move from pilot studies to scale.  

We would be interested in partnering with Professor Joseph Doyle and the MIT Sloan Health Systems Initiative.  Professor Doyle is an author on two articles related to our work.  The first is an article entitled Clinical decision support for high-cost imaging: A randomized clinical trial.  This trial dealt directly with the technology we are engaged in--clinical decision support software and its use in improving provider decision-making.  Though we have never met Professor Doyle, we are interested in better understanding the landscape of decision support software, its current uses/forms and the methods used to determine its effectiveness.  We are interested in these questions in both an international and US context; we've been approached about our software for use in rural America and with Native American groups.  To date we have largely partnered with clinical researchers and would need partners to better understand how to validate cost-effectiveness. 

The second is a paper entitled Estimating Marginal Returns to Medical Care: Evidence from at-risk newborns. This paper deals directly with the patient populations our technology targets, sick and small newborns.  The research focuses in on very low birth weight neonates, commonly defined as 1500 grams or less, a population directly addressed through our decision support software.  While we are familiar with the clinical arguments for our technology--that it is better for patients when providers follow best-practice guidelines--we are interested in learning how to make and evaluate economic arguments for neonatal decision support software.  

The NoviGuide expands and enhances healthcare workers' abilities to care for newborns.  It does this by making it easier for healthcare workers with limited formal training in the care of newborns to apply best practice guidelines to individual babies.  An anecdote may help explain NoviGuide's potential to scale newborn care.  When demonstrating the NoviGuide to a group of nurses in Uganda, a surgical nurse remarked, "I don't take care of babies, but I know how to take vital signs, place an IV and give medications.  If I got reassigned to cover for someone who was sick and had this software, I could survive the night."  This comment captured several real-world aspects of newborn care--the lack of formal training that can lead to poor care and the high staff turnover that limits the effectiveness of traditional classroom-based training methods.   

Building a highly specialized neonatal workforce in settings with extreme neonatal mortality that mirrors that seen in wealthy countries, while aspirational in the long-term, is unlikely to rapidly address crisis level neonatal deaths.  Decision support software can help bridge the gap--immediately helping frontline healthcare workers without specialized training deliver high-quality newborn care.