Detecting diabetic eye disease with machine learning

Diabetic retinopathy — an eye condition that affects people with diabetes — is the fastest growing cause of blindness, with nearly 415 million diabetic patients at risk worldwide. The disease can be treated if detected early, but if not, it can lead to irreversible blindness.

One of the most common ways to detect diabetic eye disease is to have a specialist examine pictures of the back of the eye and determine whether there are signs of the disease, and if so, how severe it is. While annual screening is recommended for all patients with diabetes, many people live in areas without easy access to specialist care. That means millions of people aren’t getting the care they need to prevent loss of vision.

A few years ago, a Google research team began studying whether machine learning could be used to screen for diabetic retinopathy (DR). Today, in the Journal of the American Medical Association, we’ve published our results: a deep learning algorithm capable of interpreting signs of DR in retinal photographs, potentially helping doctors screen more patients, especially in underserved communities with limited resources.

diabetic retinopathy
Examples of retinal photographs that are taken to screen for DR. A healthy retina can be seen on the left; the retina on the right has lesions, which are indicative of bleeding and fluid leakage in the eye.

Working with a team of doctors in India and the U.S., we created a dataset of 128,000 images and used them to train a deep neural network to detect diabetic retinopathy. We then compared our algorithm’s performance to another set of images examined by a panel of board-certified ophthalmologists. Our algorithm performs on par with the ophthalmologists, achieving both high sensitivity and specificity. For more details, see our post on the Research blog.

We’re excited by the results, but there’s a lot more to do before an algorithm like this can be used widely. For example, interpretation of a 2D retinal photograph is only one step in the process of diagnosing diabetic eye disease — in some cases, doctors use a 3D imaging technology to examine various layers of a retina in detail. Our colleagues at DeepMind are working on applying machine learning to that method. In the future, these two complementary methods might be used together to assist doctors in the diagnosis of a wide spectrum of eye diseases.

Automated, highly accurate screening methods have the potential to assist doctors in evaluating more patients and quickly routing those who need help to a specialist. We hope this study will be one of many examples to come demonstrating the ability of machine learning to help solve important problems in healthcare.

Detecting diabetic eye disease with machine learning

Diabetic retinopathy — an eye condition that affects people with diabetes — is the fastest growing cause of blindness, with nearly 415 million diabetic patients at risk worldwide. The disease can be treated if detected early, but if not, it can lead to irreversible blindness.

One of the most common ways to detect diabetic eye disease is to have a specialist examine pictures of the back of the eye and determine whether there are signs of the disease, and if so, how severe it is. While annual screening is recommended for all patients with diabetes, many people live in areas without easy access to specialist care. That means millions of people aren’t getting the care they need to prevent loss of vision.

A few years ago, a Google research team began studying whether machine learning could be used to screen for diabetic retinopathy (DR). Today, in the Journal of the American Medical Association, we’ve published our results: a deep learning algorithm capable of interpreting signs of DR in retinal photographs, potentially helping doctors screen more patients, especially in underserved communities with limited resources.

diabeticeye_900x500.jpg
Examples of retinal photographs that are taken to screen for DR. A healthy retina can be seen on the left; the retina on the right has lesions, which are indicative of bleeding and fluid leakage in the eye.

Working with a team of doctors in India and the U.S., we created a dataset of 128,000 images and used them to train a deep neural network to detect diabetic retinopathy. We then compared our algorithm’s performance to another set of images examined by a panel of board-certified ophthalmologists. Our algorithm performs on par with the ophthalmologists, achieving both high sensitivity and specificity. For more details, see our post on the Research blog.

We’re excited by the results, but there’s a lot more to do before an algorithm like this can be used widely. For example, interpretation of a 2D retinal photograph is only one step in the process of diagnosing diabetic eye disease — in some cases, doctors use a 3D imaging technology to examine various layers of a retina in detail. Our colleagues at DeepMind are working on applying machine learning to that method. In the future, these two complementary methods might be used together to assist doctors in the diagnosis of a wide spectrum of eye diseases.

Automated, highly accurate screening methods have the potential to assist doctors in evaluating more patients and quickly routing those who need help to a specialist. We hope this study will be one of many examples to come demonstrating the ability of machine learning to help solve important problems in healthcare.

Detecting diabetic eye disease with machine learning

Diabetic retinopathy — an eye condition that affects people with diabetes — is the fastest growing cause of blindness, with nearly 415 million diabetic patients at risk worldwide. The disease can be treated if detected early, but if not, it can lead to irreversible blindness.

One of the most common ways to detect diabetic eye disease is to have a specialist examine pictures of the back of the eye and determine whether there are signs of the disease, and if so, how severe it is. While annual screening is recommended for all patients with diabetes, many people live in areas without easy access to specialist care. That means millions of people aren’t getting the care they need to prevent loss of vision.

A few years ago, a Google research team began studying whether machine learning could be used to screen for diabetic retinopathy (DR). Today, in the Journal of the American Medical Association, we’ve published our results: a deep learning algorithm capable of interpreting signs of DR in retinal photographs, potentially helping doctors screen more patients, especially in underserved communities with limited resources.

diabeticeye_900x500.jpg
Examples of retinal photographs that are taken to screen for DR. A healthy retina can be seen on the left; the retina on the right has lesions, which are indicative of bleeding and fluid leakage in the eye.

Working with a team of doctors in India and the U.S., we created a dataset of 128,000 images and used them to train a deep neural network to detect diabetic retinopathy. We then compared our algorithm’s performance to another set of images examined by a panel of board-certified ophthalmologists. Our algorithm performs on par with the ophthalmologists, achieving both high sensitivity and specificity. For more details, see our post on the Research blog.

We’re excited by the results, but there’s a lot more to do before an algorithm like this can be used widely. For example, interpretation of a 2D retinal photograph is only one step in the process of diagnosing diabetic eye disease — in some cases, doctors use a 3D imaging technology to examine various layers of a retina in detail. Our colleagues at DeepMind are working on applying machine learning to that method. In the future, these two complementary methods might be used together to assist doctors in the diagnosis of a wide spectrum of eye diseases.

Automated, highly accurate screening methods have the potential to assist doctors in evaluating more patients and quickly routing those who need help to a specialist. We hope this study will be one of many examples to come demonstrating the ability of machine learning to help solve important problems in healthcare.

How Burnsville High School students create project-based learning

Editor’s note: As part of the ExploreEDU event series, schools are working with Google for Education Premier Partners to throw open their doors and invite neighboring educators to learn from their firsthand experience using Google tools to innovate and improve. To see if there is an event near you, visit the ExploreEDU site. For those that can’t join in person, we’ve asked the host schools to share their experiences and tips in a blog post. Today’s guest authors are Colleen Coleman and Anne Staum, teachers at Burnsville High School in the greater Minneapolis/St. Paul area. The district is hosting an event on December 1 and 2 with Best Buy for Education.

No matter the path our students take, the ability to guide their own learning will be an important part of their lives. At Burnsville High School, we’ve put students at the center of their education by transitioning from a lesson-based approach to a project-based one, introducing more personalized instruction and engaging students as active participants in their education.

We see technology as an essential tool in this evolution. Earlier this year, we introduced a 1:1 program schoolwide, which means all of our 2,700 students use Google Chromebooks in class and at home. We’re using tools like Chromebooks and G Suite for Education to ensure graduates are future-ready: comfortable with computing and online research, experienced in giving and receiving feedback, and confident creating content.

Here are a few ways we’re inspiring project-based learning in the classroom:

Boosting engagement with student-led content

We believe students learn best when they’re engaged as content creators. In our social studies classes, students use Chromebooks to research concepts like civic engagement and share their findings with the class. Teachers can gamify these lessons by offering extra credit to students who post the correct answer on Google Classroom, a tool that allows educators to communicate directly with students as well as manage assignments.

Our high schoolers are also using Chromebooks to work on in-depth projects and participate in learning experiences of their choice. A group of 12th graders produced a video about the founding fathers for American Government, while two juniors enrolled in a money management class recently placed second in a competitive stock market simulation, in which they virtual invested $100,000. Interdisciplinary projects like these immerse students in real-world topics while helping them develop critical skills, such as teamwork and planning.

Fostering civic inquiry

Chromebooks have helped our students better understand the world around them. Studying current events in our American government class, such as the presidential election, has encouraged students to question and critically examine their civic institutions. In a recent project, students researched competitive congressional races with their Chromebooks and organized their findings in Google Drive before presenting their findings to their classmates with Google Slides. We also used Slides to share the results of our school-wide presidential election. Without the easy-to-use tools in G Suite, it would have been much more difficult to conduct and share the results of our election simulation. These projects not only prepare students with practical knowledge that can inspire a budding passion, but also develop the analytical and presentation skills they will need to succeed as professionals.

Shortening feedback loops

Students tend to learn just as much from feedback as they do from completing an assignment.

In language arts class, for example, students need constant feedback to develop their writing. Teachers can give feedback while students are working on problems at home by collaborating in the same Google Doc. These fast feedback loops help students respond to input from teachers while it’s still fresh in their minds and fosters collaboration, another crucial real-world skill.

Today, project-based learning is inspiring our students’ curiosity and helping them identify topics they’re passionate about. We know it’s not easy to change ingrained approaches to teaching and learning, but it’s possible. It starts with encouraging students to imagine their path and direct their learning.

How Burnsville High School students create project-based learning

Editor’s note: As part of the ExploreEDU event series, schools are working with Google for Education Premier Partners to throw open their doors and invite neighboring educators to learn from their firsthand experience using Google tools to innovate and improve. To see if there is an event near you, visit the ExploreEDU site. For those that can’t join in person, we’ve asked the host schools to share their experiences and tips in a blog post. Today’s guest authors are Colleen Coleman and Anne Staum, teachers at Burnsville High School in the greater Minneapolis/St. Paul area. The district is hosting an event on December 1 and 2 with Best Buy for Education.

No matter the path our students take, the ability to guide their own learning will be an important part of their lives. At Burnsville High School, we’ve put students at the center of their education by transitioning from a lesson-based approach to a project-based one, introducing more personalized instruction and engaging students as active participants in their education.

We see technology as an essential tool in this evolution. Earlier this year, we introduced a 1:1 program schoolwide, which means all of our 2,700 students use Google Chromebooks in class and at home. We’re using tools like Chromebooks and G Suite for Education to ensure graduates are future-ready: comfortable with computing and online research, experienced in giving and receiving feedback, and confident creating content.

Here are a few ways we’re inspiring project-based learning in the classroom:

Boosting engagement with student-led content

We believe students learn best when they’re engaged as content creators. In our social studies classes, students use Chromebooks to research concepts like civic engagement and share their findings with the class. Teachers can gamify these lessons by offering extra credit to students who post the correct answer on Google Classroom, a tool that allows educators to communicate directly with students as well as manage assignments.

Our high schoolers are also using Chromebooks to work on in-depth projects and participate in learning experiences of their choice. A group of 12th graders produced a video about the founding fathers for American Government, while two juniors enrolled in a money management class recently placed second in a competitive stock market simulation, in which they virtual invested $100,000. Interdisciplinary projects like these immerse students in real-world topics while helping them develop critical skills, such as teamwork and planning.

Fostering civic inquiry

Chromebooks have helped our students better understand the world around them. Studying current events in our American government class, such as the presidential election, has encouraged students to question and critically examine their civic institutions. In a recent project, students researched competitive congressional races with their Chromebooks and organized their findings in Google Drive before presenting their findings to their classmates with Google Slides. We also used Slides to share the results of our school-wide presidential election. Without the easy-to-use tools in G Suite, it would have been much more difficult to conduct and share the results of our election simulation. These projects not only prepare students with practical knowledge that can inspire a budding passion, but also develop the analytical and presentation skills they will need to succeed as professionals.

Shortening feedback loops

Students tend to learn just as much from feedback as they do from completing an assignment.

In language arts class, for example, students need constant feedback to develop their writing. Teachers can give feedback while students are working on problems at home by collaborating in the same Google Doc. These fast feedback loops help students respond to input from teachers while it’s still fresh in their minds and fosters collaboration, another crucial real-world skill.

Today, project-based learning is inspiring our students’ curiosity and helping them identify topics they’re passionate about. We know it’s not easy to change ingrained approaches to teaching and learning, but it’s possible. It starts with encouraging students to imagine their path and direct their learning.

How Burnsville High School students create project-based learning

Editor’s note: As part of the ExploreEDU event series, schools are working with Google for Education Premier Partners to throw open their doors and invite neighboring educators to learn from their firsthand experience using Google tools to innovate and improve. To see if there is an event near you, visit the ExploreEDU site. For those that can’t join in person, we’ve asked the host schools to share their experiences and tips in a blog post. Today’s guest authors are Colleen Coleman and Anne Staum, teachers at Burnsville High School in the greater Minneapolis/St. Paul area. The district is hosting an event on December 1 and 2 with Best Buy for Education.

No matter the path our students take, the ability to guide their own learning will be an important part of their lives. At Burnsville High School, we’ve put students at the center of their education by transitioning from a lesson-based approach to a project-based one, introducing more personalized instruction and engaging students as active participants in their education.

We see technology as an essential tool in this evolution. Earlier this year, we introduced a 1:1 program schoolwide, which means all of our 2,700 students use Google Chromebooks in class and at home. We’re using tools like Chromebooks and G Suite for Education to ensure graduates are future-ready: comfortable with computing and online research, experienced in giving and receiving feedback, and confident creating content.

Here are a few ways we’re inspiring project-based learning in the classroom:

Boosting engagement with student-led content

We believe students learn best when they’re engaged as content creators. In our social studies classes, students use Chromebooks to research concepts like civic engagement and share their findings with the class. Teachers can gamify these lessons by offering extra credit to students who post the correct answer on Google Classroom, a tool that allows educators to communicate directly with students as well as manage assignments.

Our high schoolers are also using Chromebooks to work on in-depth projects and participate in learning experiences of their choice. A group of 12th graders produced a video about the founding fathers for American Government, while two juniors enrolled in a money management class recently placed second in a competitive stock market simulation, in which they virtual invested $100,000. Interdisciplinary projects like these immerse students in real-world topics while helping them develop critical skills, such as teamwork and planning.

Fostering civic inquiry

Chromebooks have helped our students better understand the world around them. Studying current events in our American government class, such as the presidential election, has encouraged students to question and critically examine their civic institutions. In a recent project, students researched competitive congressional races with their Chromebooks and organized their findings in Google Drive before presenting their findings to their classmates with Google Slides. We also used Slides to share the results of our school-wide presidential election. Without the easy-to-use tools in G Suite, it would have been much more difficult to conduct and share the results of our election simulation. These projects not only prepare students with practical knowledge that can inspire a budding passion, but also develop the analytical and presentation skills they will need to succeed as professionals.

Shortening feedback loops

Students tend to learn just as much from feedback as they do from completing an assignment.

In language arts class, for example, students need constant feedback to develop their writing. Teachers can give feedback while students are working on problems at home by collaborating in the same Google Doc. These fast feedback loops help students respond to input from teachers while it’s still fresh in their minds and fosters collaboration, another crucial real-world skill.

Today, project-based learning is inspiring our students’ curiosity and helping them identify topics they’re passionate about. We know it’s not easy to change ingrained approaches to teaching and learning, but it’s possible. It starts with encouraging students to imagine their path and direct their learning.

Save time with new custom templates in Docs, Sheets, Slides and Forms

We recently launched new tools in G Suite like Explore, Action items, and other features to help your teams save time and focus on what’s important: creating impactful work, quicker. We know time spent re-creating files in the workplace takes away from the time your team can spend collaborating and achieving results.

That’s why, today, we’re rolling out custom templates in G Suite for the Docs, Sheets, Slides and Forms files your teams use the most.

With this new feature, your team can simply submit files to shared template galleries in the Docs, Sheets, Slides and Forms home screens for your co-workers to adapt and use as needed. With these customizable templates, your teams can focus less on formatting and more on driving impact and sharing success.

submit a template gif

G Suite for Business and Education customers can require templates be approved before they appear in the gallery or restrict who can submit new templates. Admins can learn more about enabling and using custom templates on the G Suite Apps Updates Blog.

Helping schools transform, together

Over the past few years we’ve had the privilege to work closely with thousands of schools that are seeking to improve and innovate with the help of technology. Every school is different, but we’ve heard a lot of common themes from educators: that change is hard; that change is about a whole lot more than just technology; and that obstacles are often similar across districts. School leaders face many of the same challenges and opportunities, but often have limited ways to share with — and learn from — each other.

That’s why we’ve created a new hub for school leaders to share ideas, resources, and stories: The Google for Education Transformation Center.

Defining the elements of school change

We worked alongside U.S. education leaders from across all 50 states to create a transformation framework that can help guide schools on their journeys to improve education through innovation and technology. As we talked with these leaders, they emphasized seven areas that are critical to consider when tackling the change process:

  • Vision – School change only happens when there is a strong vision at the start. When a school has a clear vision, it means the leader has ensured that the school and wider community are working together toward shared goals for the future.

  • Learning – School leaders empower their teams to create a set of instructional practices, curricula, assessments, and learning experiences that put students at the center – that engage learners deeply and meet their individual and collective needs.

  • Culture – Successful school leaders create structures, rituals, stories, and symbols that foster a culture of innovation and encourage people to learn from failure and success.

  • Technology – Technology is only one enabler of school change, but it’s a critical part. School leaders find, test, and gain their team’s support for the right technology (tools and processes) to meet the school’s vision.

  • Professional Development – Teachers have a lot on their plates. School leaders provide educators with effective professional development and ongoing coaching focused on applying tools and practices to meet student needs.

  • Funding & Sustainability – School leaders create a sustainable budget, identify a range of funding sources, and seek savings and reallocation opportunities that align directly to student goals.

  • Community – Schools serve diverse communities made up of parents, families, businesses, government, nonprofits, and residents. Throughout all stages of the transformation process, leaders ensure these partners support the school and the vision.

Successful school leaders considered all of these elements simultaneously. Michael Lubelfeld, Superintendent of Deerfield Public School District 109 said, “Narrow school improvement models don’t work. At Deerfield we’ve adopted a much more holistic school improvement planning process that utilizes the power and collective capacity of student, staff, parent, and community voice.”

Share what’s worked (and what hasn’t)

Change is often a gradual process, but we hope that helping school leaders share ideas will accelerate their school transformations. I’m excited we can provide a platform where they can learn from each other.

Nick Polyak, the Superintendent of Leyden High School District 212 in Illinois explained how important it has been for his district to learn from others. He said “There are great ideas and great programs all over the world.  When we are willing to share and learn from one another, all of our students benefit. We have benefited from visiting other districts and adapting their programs for our schools.”

There are great ideas and great programs all over the world. When we are willing to share and learn from one another, all of our students benefit.

Nick Polyak

Superintendent, Leyden HS District 212

To continue to improve the Google for Education Transformation Center, we’re asking you, our community of education leaders, to share your own resources. These might include stories (like how you worked with your team to develop a strong vision), templates (like your district’s schedule of teacher-led professional development), or ideas (like how you raised funds to get WiFi on school buses). We’ll work with educators to review submissions as we continue to build out the site. Submit by November 23rd to have your ideas considered for inclusion in an upcoming series of blog posts around our Education on Air online conference.

[edu] transformation gif

Google Science Fair 2016: a look back, and a giant leap ahead

Editor’s note: Today’s post comes from our first-ever Google Science Fair Grand Prize winner in 2011, Shree Bose. Back for the annual Science Fair five years later, here she shares her own story as a glimpse of what’s in store for this year’s winners. Thanks to our partners at Lego Education, National Geographic, Scientific American and Virgin Galactic for another great year of Google Science Fair. In case you missed it, you can still catch the livestream!

Five years ago to the day, I was a finalist in the first-ever Google Science Fair — a program where any student 13-18 is invited to solve the world’s biggest challenges through science and technology. I was fascinated by the peculiar ways cancer cells process energy and wondered if we might be able to target those processes. So, the idea behind my project was to study AMP kinase, an energy protein, to understand its importance in the way ovarian cancer cells develop resistance to drugs. I was 17 when I won the Grand Prize, and my life hasn’t been the same since.

Today 20 of the world’s brightest young scientists have that same chance ahead of them, and I am so excited and grateful to be here with them in Mountain View to re-live that experience.

The 2016 Google Science Fair finalists 

Meeting President Obama(!)
Official White House photo by Pete Souza

For me, the Google Science Fair took my passion for science and gave me a global platform to share it with the world. I went on to do my undergraduate studies at Harvard University, majoring in Molecular and Cellular Biology with a minor in Global Health and Health Policy. My interest in using new metabolic tools to study cancer has led me to Duke University School of Medicine, where I’m currently pursuing an MD/PhD and looking forward to a career that brings together clinical medicine and basic science. But five years ago, it was the Google Science Fair that first provided me with the platform to share my ideas, unlocking doors to some incredible, once-in-a-lifetime opportunities: meeting President Obama, speaking at TEDx events all over the world and being included on Glamour magazine’s list of Top 10 College Women.

Back at Google today, I witnessed more than 500 students from local Bay Area schools — a majority that qualified for Title I funding — as they asked questions, found inspiration, and saw science and engineering in new, unexpected ways.
Those students had a chance to talk to the finalists, from 9 countries, who are working on things like better diagnosing cancer, fighting drought with fruit, training robotic hands, developing compostable feminine sanitary products and so much more. (I also have to point out, this is the first year a majority of the finalists are female, which makes me especially proud!)

My sincere congratulations to all of this year’s Google Science Fair finalists and winners. You might “just” be teenagers, but you’re also amazing researchers, entrepreneurs, technologists and explorers who are challenging themselves — and all of the rest of us — to make things better. To quote last year’s Google Science Fair winner (and my friend) Olivia Hallisey who is also back here this week as a judge:

“Every one of us, no matter our age or background, can make a difference. But change doesn’t happen overnight, and it often starts with a question. So look at the world around you, and challenge yourself to make something better.”

As for me, the Science Fair gave me the confidence to continue asking questions, developing a passion for science and engineering, and even to co-found Piper, a company focused on developing electrical engineering kits for kids to learn the basics of building hardware. I just can’t wait to see where you will be five years from today.

And finally, announcing the 2016 Google Science Fair winners:

The 2016 Google Science Fair winners!

Posted by Shree Bose, 2011 Google Science Fair winner https://3.bp.blogspot.com/-6FNNcKSgLiM/V-rrBkuD8YI/AAAAAAAATF0/tvOg4g4kS4c6TBR0QiALygZnB_GEfJkbACLcB/s1600/_MG_3901.jpg Shree Bose 2011 Google Science Fair winner <!– INSTRUCTIONS Enter info below to be used in google.com/about site blog syndication. Leave elements empty if there is no valid data. Example: https://2.bp.blogspot.com/-gJ7uwee_8mk/V-rj1pI27XI/AAAAAAAATFk/ugK_9VoNOhksj36jpWDLJULiyqoFYyWawCLcB/s1600/_MG_3901.jpg –>