Chapter 1
Align purpose, strategy and leadership
When planning an analytics initiative, organizations will benefit from a clear ambition, strong leadership and a road map for change.
1. State the ambition
Analytics initiatives should begin with a clear sense of purpose, grounded firmly in the service-user need or problem that needs tackling. Sara Vaezy, Chief Digital Strategy and Business Development Officer at Providence, believes that correctly identifying the problem up front is the key. “We’ve always taken a problem-first approach, or an opportunity-first approach, and that has made us laser-focused on all of the other supportive things, including the data and the analytics, that we need.”
Ultimately, you’ve got to bring that human, lived experience into any model. Talk to the data subjects first, before you go away and do any kind of machine learning or bring any data together.
It’s also important to make sure that service-user needs are center-stage in designing the solution. At LBBD, Head of Insight and Innovation Pye Nyunt says: “Ultimately, you’ve got to bring that human, lived experience into any model. Talk to the data subjects first before you go away and do any kind of machine learning or bring any data together.”
2. Ensure leaders drive change
All of the case study organizations had one thing in common: strong support from senior leadership who are open to new approaches. “That’s how we can do the things we do, because we’ve got senior management that think outside of the box,” said LBBD Service Manager Gill Wilson. “They’re not constrained by the way things were done before, they’re very forward-looking.”
A powerful way to win leaders’ buy-in is to give a voice to service users at the outset. Director of Person-Centered Care at Ontario Health (Cancer Care Ontario) Colleen Fox says: “It’s very different when you hear a patient come up and tell their story … When they say either ‘This could have benefited me’ or ‘This system was in place and this is how it helped me,’ that’s been very powerful for us, for leadership buy-in.”
Once leaders are on board, it is important that they become champions for the project, to explain the need for change and transmit enthusiasm through the whole organization. At Ontario Health, Regional Oncology Lead for Patient-Centered Care Christine Peters describes one particular figure from the initial phase of the project: “We had a provincial head who oversaw a few programs, most prominently nursing. … She was a rallying force, not only internally in the organization to get something set up, but also more forward-facing with the hospitals who are delivering cancer services.” Ontario Health now has clinicians and leaders throughout the organization who champion the program and continue to promote the vision.
3. Outline clear goals and a roadmap
Clear goals, a roadmap and a resourcing plan should underpin the analytics initiative. Derek Streat, CEO of Providence’s DexCare, says: “Lots of initiatives in health systems die because you don’t have an organization that is pointing at a mountain on the horizon saying, ‘That’s the mountain we’re going to climb, and here’s how we’re going to resource it … and here is how we’re going to staff it.’”
To support organizations as they begin their journey, the full report from the study includes a diagnostic for assessing readiness for embracing analytics solutions, including some key questions to address internally and with potential partners.
4. Tailor the analytics program to local conditions
Analytics programs should be tailored to the local context. In Hillsborough County, support and funding from the county allowed the intervention to be driven locally, ensuring that the Family Preservation and Assessment System would encapsulate local needs. Christopher Card, Chief of Community-Based Care at Eckerd Connects, the private contractor that delivers child and juvenile care services in the county, says it is important to work with the organizations that are already on the ground and that families know. This ensures that the solution can build on the trust these organizations already have in the community.
Although Hillsborough County leveraged insights and overall platform structure based on EY’s earlier work in Australia, everyone involved was at pains to make sure it genuinely reflected local realities, rather than imposing a pre-existing template. Heather Cazzola, Director of Diversion and Prevention at Eckerd Connects, says: “The framework and everything we’ve done is very different from the platform in Australia. … They’ve really been able to adapt to our specific county and state needs.”
5. Promote collaboration
Collaboration – both internal and external – is key. Ontario Health brought together many disciplines from within the organization: the legal team, the privacy team, the technical services team and the data and analytics team. Colleen Fox says: “A lot of internal teams really had to come together in order to roll out PROMs and ensure it continues to be sustainable within our system.”
External collaboration comes into play when data sets are owned by different organizations, or when different organizations are responsible for delivering services. All stakeholders should be involved from the early stages of planning, so they have a clear sense of the overall vision and objectives, the role their organization plays in achieving these, and how data and analytics will be used to achieve the objectives. Stakeholders must agree on shared measurements of success against which all will be held accountable.
Hillsborough, for example, has created a shared vision of the future of child services across the whole area. Christopher Card says: “You need to collaborate, bring people together and take the time to get them on board with the shared vision. Then include them in the execution to the extent possible, so that it’s an execution at the local level, so they stay bought in and they stay aware and they stay on your team getting this done.” This shared vision, and a new culture of collaboration, has already started to break down the siloed ways in which social services are delivered. The team also believes that it reduces the risk of a project being abandoned if its main champion leaves the organization.
Chapter 2
Create the business case for adequate funding
Securing funding will always be a challenge. It’s vital to identify long-term savings, target quick wins and explore commercial partnerships.
1. Identify long-term savings
Analytics solutions require substantial investment and funding is not always readily available, particularly when governments are constrained by short-term budget cycles. To address this, organizations need to demonstrate likely long-term benefits, including the potential for lower costs to be incurred downstream. Ramin Kouzehkanani, Chief Information and Innovation Administrator for Hillsborough County, says: “I believe the cost of not doing it is ultimately greater than the cost of doing it, especially when [organizations] consider the prevention outcomes and the maintenance and operational cost of their legacy systems.”
In the UK, councils like Barking and Dagenham are targeting preventative services where possible, particularly in areas such as social care, because it reduces costs and improves outcomes for service users. LBBD has found that insights generated through One View have helped to reduce the use of temporary accommodation, leading to significant savings. By supporting citizens to stay out of debt and remain in their own homes, One View also helps maintain council tax revenues, in turn protecting other services.
I believe the cost of not doing it is ultimately greater than the cost of doing it, especially when [government agencies] consider the prevention outcomes and the maintenance and operational cost of their legacy systems.
2. Target quick wins
Some organizations focus on achieving quick wins that can prove the benefits and overcome the challenge of short political funding cycles. Rhodri Rowlands, Head of Programs for Community Solutions at LBBD, suggests: “Seek to start small and … demonstrate the value of that. Use that as a mechanism to bring others on the journey.” After two years, having shown the value of early data-driven solutions both internally and externally, LBBD was ready to partner with an external organization, enabling provision of a large-scale platform supporting more than 400 staff.
3. Explore commercial partnerships
For some, commercial partnerships have proved important in finding more cost-effective solutions. In Israel, Sheba uses creative mechanisms to engage clinicians, researchers and data scientists in its work. Through commercial partnerships, it has found a way to pair its data and clinical knowledge with digital expertise in industry, allowing it to develop cutting-edge digital innovations without the backing of multi-million-dollar budgets. The former Head of the Big Data and AI Hub Dr Nathalie Bloch says: “With a lot of collaboration … and negotiations where both sides win the situation, with the industry, then we are able to create very nice projects without a high budget.”
Chapter 3
Build public confidence in the use of data
To overcome service user wariness, organizations must demonstrate the benefits of analytics projects.
1. Build trust in data use
EY research has shown the general public remains skeptical and cautious regarding the use of their data. In particular, they often feel uneasy about data sources being combined to develop a fuller picture. Our Connected Citizens survey of 12,000 citizens in 12 countries revealed that around 40% are against the sharing of their personal data. As such, trust building can no longer be an afterthought. The public benefit of data sharing needs to be proven and demonstrated.
Service users’ acceptance of the use of their data is central to the success of analytics programs. Ontario Health’s program depends on patients taking the time to fill out the screenings about their own health ahead of their consultations. The organization is acutely aware that, without patient buy-in, no data will be collected. Joanne MacPhail, a Patient and Family Advisors representative, says: “There is no initiative, no committee that does any work without at least two patients and caregivers at the table.”
Meanwhile, LBBD acknowledges the controversy over algorithms. Its communication strategy emphasizes that One View does not make decisions about people’s lives – rather, the platform presents aggregated information to a professional who can interrogate the data to make more informed decisions.
Building service user trust can also be facilitated by using “familiar faces” such as clinicians or case workers to explain the benefits. At Hillsborough, the team believes this is a fundamental reason why change should be driven by organizations that are already active in the community.
There is no initiative, no committee that does any work without at least two patients and caregivers at the table.
2. Be transparent
Organizations need to specify what data is being used or shared, with whom, and for what purpose. This allows the public to see the benefits and make their own judgments about the value of data sharing and analytics.
LBBD uses external scrutiny to help build trust. The organization encourages external experts to review and evaluate its services; for example, researchers from the Ada Lovelace Institute, an independent body focused on ensuring that data and AI serve society, are studying how LBBD staff use One View and are hoping to learn from these independent observations. LBBD is also developing a Data Ethics and Transparency Charter that will provide clarity on the issues that tend to cause most anxiety to the public.
Chapter 4
Create robust data quality and governance frameworks
Safeguards are needed to improve the richness and relevance of data, while strict rules should be adopted to govern who has access.
1. Ensure data quality
Successful organizations take steps to improve the quality, richness and relevance of their available data so it is fit for analysis. This involves putting safeguards in place to ensure data relevance and quality, including data cleansing to remove errors, duplications and conflicting information. It’s important not to underestimate how long it can take to convert information into reliable, consistent and analyzable data sets.
At LBBD, data gets pulled from multiple sources. Pye Nyunt says: “Data quality was a challenge. … The good thing with data cleansing is although it’s painful to begin with, once you’ve done it, you’ve built a script that can make sure that it keeps it consistent.”
Sigal Sina, Chief Data Scientist at the Big Data and AI Hub, Sheba Medical Center, says her team’s main challenge is collecting and pre-processing the data that sit in more than 40 siloed repositories across the organization, then combining it into one data lake.
The good thing with data cleansing is, although it’s painful to begin with, once you’ve done it, you’ve built a script that can make sure that it keeps it consistent.
2. Implement robust governance
The proliferation in the scale and availability of data raises further questions about who should have access to data, what they should be allowed to do with it, and what rules and regulations should be in place to protect it. A robust information governance (IG) approach is therefore fundamental to the success of any analytics solution.
Ontario Health has the status of prescribed entity, allowing it to use personal health information for planning purposes, provided it adheres to strict data management and security protocols. This includes a robust information governance structure and accountability agreements with each hospital specifying what data fields are collected and why. Whenever a new PROM is introduced, it goes through a risk assessment, a privacy impact assessment, and several other checkpoints and procedures.
For Gill Wilson, Service Manager at LBBD, the IG approach gives her confidence that she’s using the data appropriately. She says: “One View has got a very, very good structure to it, that embeds the Data Protection Act, GDPR [the EU’s General Data Protection Regulation], so that you know whenever you go in, you only get access to what is at your level.”
In some organizations, a lot of the data shared is anonymized. Data sets can be linked in a central way that protects the identity of individuals. This technique is useful when data can be aggregated and the details of the individual never need to be known, for example to predict where in a local authority the largest demand for a service will come from.
However, if the details of the individual may need to be known at some point, for example to provide safeguarding to a child, anonymizing the data before use is not an option. In this instance, pseudonymization can be explored as a way to increase the privacy of individuals while allowing re-identification if required.
Chapter 5
Support staff adoption and buy-in
No analytics project can succeed unless staff embrace it. Good design, support and training are key.
1. Engage staff through user-centered design
Many frontline workers feel that digital technologies detract from their interactions with their clients and add an extra layer of work to their daily routine. To ensure that staff can easily embed analytics into their day-to-day work, it is vital to design solutions around their needs and preferences.
This product is really a co-development effort with others in critical business and operation lines.
Nisim Rahman, Data Architect and Team Lead at the Big Data and AI Hub, Sheba Medical Center, says: "To build a system that will help the doctors, you need to listen to the doctors, to understand exactly what they want.” This clinician involvement in design and development of innovations also helps to create buy-in from staff and facilitates the adoption of the innovation in practice.
The same is true at Providence, where consultation with different service lines ensured the final product fitted seamlessly into their workflow. Group Vice President, Chief Products Officer at Digital Innovations Maryam Gholami says: “This product is really a co-development effort with others in critical business and operation lines.”
2. Invest in training
Providing front-line practitioners with training and support aids staff buy-in and effective use of analytics systems. Ideally, this training will involve those members of staff that have co-developed the solution with the technical team.
At LBBD, for example, training has been largely shaped by One View Champions, front-line staff who guide practitioners on how to access and use the tool. Importantly, the training covers the ethics and IG questions related to data access. Katy Brown, Programs and Strategy Officer says “Quite understandably, that was … the first question that almost everyone asked us in a training session: are we allowed to access this information? We spend quite a lot of time doing training with all staff.” The team also provides supporting material for future reference, FAQ documents, and refresher training.
For other organizations, staff training was seen as a missed opportunity. Colleen Fox of Ontario Health feels that mandatory, standardized training at the outset could have been a good idea. “I’ve always wondered if that could have helped, even in the early days, to set us up for success. … It’s something that I would do differently if we were starting over.”
3. Highlight the benefits and address staff concerns
As Godfred Boahen, formerly Policy, Research and Practice Improvement Lead, Association of British Social Workers observed, “There is a crisis of cynicism around digital technologies in the social services sector because the benefits are not articulated clearly.”
It is vital that staff can see how analytics can benefit them in their own roles, for example, by improving decision-making. In Hillsborough County, one of the goals of the Family Preservation and Assessment System was to provide child protective investigators with more information about their cases. This drove the team to ensure that input from potential users is incorporated into the system early on. So far, they have seen a very positive acceptance of the system.
Related article
“There is a crisis of cynicism around digital technologies in the social services sector because the benefits are not articulated clearly.
Importantly, organizations can reassure staff that the tools will be used to enhance rather than replace their professional judgment. At Ontario Health, clinical leads for PROMs visit individual sites to address any staff concerns. Dr. Natalie Coburn says: “A lot of times it’s just a matter of … saying, "Okay, well, we’re not collecting this data for hoarding purposes. … We have randomized controlled trial data that shows that if you intervene on these symptoms, patients live longer.’”
Meanwhile, at LBBD, there is a One View Mailbox where staff can get in touch with the project team confidentially to voice any questions or concerns.
Chapter 6
Encourage continual innovation to embed change
The environment for continual innovation must be consciously designed and nurtured to ensure progress is maintained.
1. Evolve through evaluation and feedback
To ensure solutions deliver results and keep pace with the organization’s evolving needs, a mechanism for continuous feedback and improvement is required. On Providence’s DexCare project, staff can provide continuous feedback on the tool and are also looped in for future developments. Periodically, feedback is sought from clinical staff and providers on specific launches, ideas and pilots.
At LBBD, feedback from front-line staff drives improvements to One View. Service Manager Jill Gallagher says: "You need to get feedback from staff because they are the people using it day in and day out, and they are very honest. Front-line staff give us some great narratives around what’s going on for them and how the needs have changed from the people they’re serving. We are able to take that information and then adapt One View to meet that need.”
Another tactic is to recognize that failure is an intrinsic part of eventual success. Dr. Nathalie Bloch of Sheba says the organization continuously evaluates its analytics solutions against the original goal and revisits them if they are not working. She says: "Be ready to fail, fail, fail, fail, fail, and only then to maybe succeed. … You need to be very patient to see results.”
2. Incentivize fresh thinking
To fully leverage the potential of data and analytics, organizations will benefit from fostering an innovative and entrepreneurial culture. One way to do this is by rewarding individuals when they try something new or think beyond the traditional ways of planning and executing programs. During 2020, Sheba handed out 10 grants of up to US$50,000 each, to staff initiatives in big data and AI. Sheba’s Dr. Robert Klempfner says the culture is one of “curiosity and entrepreneurship, innovation [and thinking] ‘you can’t just complain, you need to try and do something about it.’”
3. Harness technical capabilities
Key to success is the ability to build blended teams that combine tech expertise from outside with in-house experts who really understand service users. In the US, Providence’s Digital Innovation Group (DIG), which oversaw the development of DexCare, is formed of longstanding employees who understand the organization, leavened with external hires who have extensive experience of working for Big Tech companies. David McAughan, Executive Director of Express Care and Line of Business Leader, says: “I think that blending of capabilities has really set them up to being great partners.”
Sometimes it is difficult for local authorities to hire top talent from tech companies. In these cases, partnerships can help to fill the gap. For example, LBBD partnered with Big Data firm Xantura to deliver the expertise it needed.
4. Foster a network
Organizations can consider sharing their experiences with a diverse range of global partners, creating a community where organizations learn from each other, rather than having to start from scratch when developing new solutions or entering negotiations with prospective partners. Sheba’s innovation hub engages with the wider ecosystem, such as data companies and global academic institutions, to design, develop and deploy new projects. Dr. Robert Klempfner says: “The ecosystem is crucial. … It’s important to have partners across the world, where you can collaborate, you can cross-validate your projects.”
Related article
Conclusion
Truly innovative organizations understand that the analytics journey is never completed. As each data-driven project comes to fruition, it reveals new horizons for exploration and progress.
All five of the organizations in our study have ambitious plans. Providence aims to advance DexCare’s capabilities by introducing predictive analytics. In Hillsborough County, there are plans to establish a community-wide aggregated data lake, identify further data sets to be included, and develop more advanced functions such as real-time data visualization. At Sheba, the future of improved care encompasses integrating data sets from beyond the hospital. At Ontario Health, the organization believes it is only just realizing the potential benefits that data can bring in terms of supporting patients. And at LBBD, plans include improving the accuracy of the data, adding new data sources and becoming more data-led as an organization.
For those that are only at the start of their analytics journey, the road ahead may seem daunting. But the five organizations in our study prove the power of making a start, thinking carefully about a roadmap, pursuing early wins to demonstrate expected benefits, and breaking down siloes to reach a unified mission. They all set a clear objective and advanced one step at a time, creating a template for success that can be emulated elsewhere.
Summary
During the pandemic, health and human services organizations around the world have embraced digital technology, which in turn has generated a wealth of data. As the crisis recedes, there is an opportunity to develop analytics initiatives that harness this data and use it to improve citizens’ lives. This article provides actionable insights from pioneering organizations that have already launched successful analytics projects.
