Array
(
[body] => Array
(
[#theme] => field
[#weight] => 0
[#title] => Body
[#access] => 1
[#label_display] => hidden
[#view_mode] => teaser
[#language] => und
[#field_name] => body
[#field_type] => text_with_summary
[#field_translatable] => 0
[#entity_type] => node
[#bundle] => box_lancio_news
[#object] => stdClass Object
(
[vid] => 500796
[uid] => 26499
[title] => Unipd alumni develop AI to detect diabetes early
[log] =>
[status] => 1
[comment] => 0
[promote] => 1
[sticky] => 0
[nid] => 120490
[type] => box_lancio_news
[language] => it
[created] => 1756200101
[changed] => 1757324322
[tnid] => 0
[translate] => 0
[revision_timestamp] => 1757324322
[revision_uid] => 2032
[body] => Array
(
[und] => Array
(
[0] => Array
(
[value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[summary] =>
[format] => 2
[safe_value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[safe_summary] =>
)
)
)
[field_date_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_etichetta_box_lancio_news] => Array
(
)
[field_img_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[fid] => 142067
[uid] => 26499
[filename] => ai_salute.jpg
[uri] => public://ai_salute.jpg
[filemime] => image/jpeg
[filesize] => 229888
[status] => 1
[timestamp] => 1756199852
[type] => image
[field_file_image_alt_text] => Array
(
)
[field_file_image_title_text] => Array
(
)
[field_folder] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2048
)
)
)
[metadata] => Array
(
[height] => 1322
[width] => 2519
)
[height] => 1322
[width] => 2519
[alt] => digital screen with images of the human brain
[title] =>
)
)
)
[field_link_alla_news] => Array
(
)
[field_link_esterno_news] => Array
(
)
[field_pagina_associata] => Array
(
)
[field_link_etichetta] => Array
(
)
[field_abstract_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
[format] =>
[safe_value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
)
)
)
[field_allegato_news] => Array
(
)
[field_categorie_news] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2296
)
)
)
[field_pub_date] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[value2] => 2026-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_layout_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => single
)
)
)
[field_testo_opzionale_news] => Array
(
)
[field_url_en_page] => Array
(
[und] => Array
(
[0] => Array
(
[value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
[format] =>
[safe_value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
)
)
)
[field_url_en_page_label] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Italian version
[format] =>
[safe_value] => Italian version
)
)
)
[path] => Array
(
[pathauto] => 0
)
[name] => rossella.vezzosi
[picture] => 0
[data] => b:0;
[num_revisions] => 5
[current_revision_id] => 500796
[is_current] => 1
[is_pending] =>
[revision_moderation] =>
[entity_view_prepared] => 1
)
[#items] => Array
(
[0] => Array
(
[value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[summary] =>
[format] => 2
[safe_value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[safe_summary] =>
)
)
[#formatter] => text_summary_or_trimmed
[0] => Array
(
[#markup] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
)
)
[field_img_box_lancio_news] => Array
(
[#theme] => field
[#weight] => 0
[#title] => Immagine
[#access] => 1
[#label_display] => above
[#view_mode] => teaser
[#language] => und
[#field_name] => field_img_box_lancio_news
[#field_type] => image
[#field_translatable] => 0
[#entity_type] => node
[#bundle] => box_lancio_news
[#object] => stdClass Object
(
[vid] => 500796
[uid] => 26499
[title] => Unipd alumni develop AI to detect diabetes early
[log] =>
[status] => 1
[comment] => 0
[promote] => 1
[sticky] => 0
[nid] => 120490
[type] => box_lancio_news
[language] => it
[created] => 1756200101
[changed] => 1757324322
[tnid] => 0
[translate] => 0
[revision_timestamp] => 1757324322
[revision_uid] => 2032
[body] => Array
(
[und] => Array
(
[0] => Array
(
[value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[summary] =>
[format] => 2
[safe_value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[safe_summary] =>
)
)
)
[field_date_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_etichetta_box_lancio_news] => Array
(
)
[field_img_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[fid] => 142067
[uid] => 26499
[filename] => ai_salute.jpg
[uri] => public://ai_salute.jpg
[filemime] => image/jpeg
[filesize] => 229888
[status] => 1
[timestamp] => 1756199852
[type] => image
[field_file_image_alt_text] => Array
(
)
[field_file_image_title_text] => Array
(
)
[field_folder] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2048
)
)
)
[metadata] => Array
(
[height] => 1322
[width] => 2519
)
[height] => 1322
[width] => 2519
[alt] => digital screen with images of the human brain
[title] =>
)
)
)
[field_link_alla_news] => Array
(
)
[field_link_esterno_news] => Array
(
)
[field_pagina_associata] => Array
(
)
[field_link_etichetta] => Array
(
)
[field_abstract_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
[format] =>
[safe_value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
)
)
)
[field_allegato_news] => Array
(
)
[field_categorie_news] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2296
)
)
)
[field_pub_date] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[value2] => 2026-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_layout_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => single
)
)
)
[field_testo_opzionale_news] => Array
(
)
[field_url_en_page] => Array
(
[und] => Array
(
[0] => Array
(
[value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
[format] =>
[safe_value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
)
)
)
[field_url_en_page_label] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Italian version
[format] =>
[safe_value] => Italian version
)
)
)
[path] => Array
(
[pathauto] => 0
)
[name] => rossella.vezzosi
[picture] => 0
[data] => b:0;
[num_revisions] => 5
[current_revision_id] => 500796
[is_current] => 1
[is_pending] =>
[revision_moderation] =>
[entity_view_prepared] => 1
)
[#items] => Array
(
[0] => Array
(
[fid] => 142067
[uid] => 26499
[filename] => ai_salute.jpg
[uri] => public://ai_salute.jpg
[filemime] => image/jpeg
[filesize] => 229888
[status] => 1
[timestamp] => 1756199852
[type] => image
[field_file_image_alt_text] => Array
(
)
[field_file_image_title_text] => Array
(
)
[field_folder] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2048
)
)
)
[metadata] => Array
(
[height] => 1322
[width] => 2519
)
[height] => 1322
[width] => 2519
[alt] => digital screen with images of the human brain
[title] =>
)
)
[#formatter] => image
[0] => Array
(
[#theme] => image_formatter
[#item] => Array
(
[fid] => 142067
[uid] => 26499
[filename] => ai_salute.jpg
[uri] => public://ai_salute.jpg
[filemime] => image/jpeg
[filesize] => 229888
[status] => 1
[timestamp] => 1756199852
[type] => image
[field_file_image_alt_text] => Array
(
)
[field_file_image_title_text] => Array
(
)
[field_folder] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2048
)
)
)
[metadata] => Array
(
[height] => 1322
[width] => 2519
)
[height] => 1322
[width] => 2519
[alt] => digital screen with images of the human brain
[title] =>
)
[#image_style] =>
[#path] =>
)
)
[field_abstract_news] => Array
(
[#theme] => field
[#weight] => 0
[#title] => Abstract
[#access] => 1
[#label_display] => above
[#view_mode] => teaser
[#language] => und
[#field_name] => field_abstract_news
[#field_type] => text_long
[#field_translatable] => 0
[#entity_type] => node
[#bundle] => box_lancio_news
[#object] => stdClass Object
(
[vid] => 500796
[uid] => 26499
[title] => Unipd alumni develop AI to detect diabetes early
[log] =>
[status] => 1
[comment] => 0
[promote] => 1
[sticky] => 0
[nid] => 120490
[type] => box_lancio_news
[language] => it
[created] => 1756200101
[changed] => 1757324322
[tnid] => 0
[translate] => 0
[revision_timestamp] => 1757324322
[revision_uid] => 2032
[body] => Array
(
[und] => Array
(
[0] => Array
(
[value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[summary] =>
[format] => 2
[safe_value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[safe_summary] =>
)
)
)
[field_date_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_etichetta_box_lancio_news] => Array
(
)
[field_img_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[fid] => 142067
[uid] => 26499
[filename] => ai_salute.jpg
[uri] => public://ai_salute.jpg
[filemime] => image/jpeg
[filesize] => 229888
[status] => 1
[timestamp] => 1756199852
[type] => image
[field_file_image_alt_text] => Array
(
)
[field_file_image_title_text] => Array
(
)
[field_folder] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2048
)
)
)
[metadata] => Array
(
[height] => 1322
[width] => 2519
)
[height] => 1322
[width] => 2519
[alt] => digital screen with images of the human brain
[title] =>
)
)
)
[field_link_alla_news] => Array
(
)
[field_link_esterno_news] => Array
(
)
[field_pagina_associata] => Array
(
)
[field_link_etichetta] => Array
(
)
[field_abstract_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
[format] =>
[safe_value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
)
)
)
[field_allegato_news] => Array
(
)
[field_categorie_news] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2296
)
)
)
[field_pub_date] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[value2] => 2026-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_layout_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => single
)
)
)
[field_testo_opzionale_news] => Array
(
)
[field_url_en_page] => Array
(
[und] => Array
(
[0] => Array
(
[value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
[format] =>
[safe_value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
)
)
)
[field_url_en_page_label] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Italian version
[format] =>
[safe_value] => Italian version
)
)
)
[path] => Array
(
[pathauto] => 0
)
[name] => rossella.vezzosi
[picture] => 0
[data] => b:0;
[num_revisions] => 5
[current_revision_id] => 500796
[is_current] => 1
[is_pending] =>
[revision_moderation] =>
[entity_view_prepared] => 1
)
[#items] => Array
(
[0] => Array
(
[value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
[format] =>
[safe_value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
)
)
[#formatter] => text_default
[0] => Array
(
[#markup] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
)
)
[links] => Array
(
[#theme] => links__node
[#pre_render] => Array
(
[0] => drupal_pre_render_links
)
[#attributes] => Array
(
[class] => Array
(
[0] => links
[1] => inline
)
)
[node] => Array
(
[#theme] => links__node__node
[#links] => Array
(
[node-readmore] => Array
(
[title] => Read more
about Unipd alumni develop AI to detect diabetes early
[href] => node/120490
[html] => 1
[attributes] => Array
(
[rel] => tag
[title] => Unipd alumni develop AI to detect diabetes early
)
)
)
[#attributes] => Array
(
[class] => Array
(
[0] => links
[1] => inline
)
)
)
)
[field_date_box_lancio_news] => Array
(
[#theme] => field
[#weight] => 1
[#title] => Data
[#access] => 1
[#label_display] => above
[#view_mode] => teaser
[#language] => und
[#field_name] => field_date_box_lancio_news
[#field_type] => date
[#field_translatable] => 0
[#entity_type] => node
[#bundle] => box_lancio_news
[#object] => stdClass Object
(
[vid] => 500796
[uid] => 26499
[title] => Unipd alumni develop AI to detect diabetes early
[log] =>
[status] => 1
[comment] => 0
[promote] => 1
[sticky] => 0
[nid] => 120490
[type] => box_lancio_news
[language] => it
[created] => 1756200101
[changed] => 1757324322
[tnid] => 0
[translate] => 0
[revision_timestamp] => 1757324322
[revision_uid] => 2032
[body] => Array
(
[und] => Array
(
[0] => Array
(
[value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[summary] =>
[format] => 2
[safe_value] =>
A team of researchers from the Scripps Research Translational Institute, including alumni from the University of Padua, has developed an artificial intelligence model that can early detect the risk of diabetes by analysing glucose spikes through wearable sensors. Traditionally, diabetes and prediabetes are diagnosed using the HbA1c test, which measures average blood glucose levels over the past few months. However, this test does not predict who is at risk of developing diabetes.
Mattia Carletti, the first author, Matteo Gadaleta, in charge of data processing, and Giorgio Quer, senior co-author and corresponding author, conducted the study at Scripps Research. Riccardo Miotto works at Tempus AI, the sponsor of the study, and managed the collaboration. All four researchers come from the University of Padua, where they completed their PhDs in the Department of Information Engineering. In this study, they discovered that artificial intelligence can use a combination of other data — including real-time glucose levels monitored by wearable devices — to provide a more accurate view of diabetes risk.
The new model utilises data from continuous glucose monitors (CGM), information on gut microbiome, diet, physical activity, and genetics, offering a more detailed view of diabetes risk. "We have shown that two people with the same HbA1c value can have very different underlying risk profiles," says Giorgio Quer, director of Artificial Intelligence and lecturer in Digital Medicine at Scripps Research. "By analysing more data, such as how long it takes for glucose spikes to return to normal, what happens to glucose during the night, dietary intake, and even gut activity, we can begin to distinguish who is on a fast track towards diabetes and who is not."
Using this information,, the researchers trained an artificial intelligence model to distinguish between people with type 2 diabetes and healthy individuals. One of the clearest signals of diabetes risk identified was the time it took for a glucose spike to return to normal values. In people with type 2 diabetes, it often took 100 minutes or more for blood glucose to drop after a spike, whereas in healthy individuals it returned to baseline values much more quickly. The study also found that a more diverse gut microbiome and higher levels of physical activity were associated with better glucose control, while a higher resting heart rate was linked to diabetes.
The study, published in "Nature Medicine," involved over 1,000 participants in the United States through a remote clinical trial, where participants used CGM devices, recorded meals, monitored physical activity, and sent biological samples for analysis. The results show that a diverse gut microbiome and high levels of physical activity are associated with better glucose control.
The AI model demonstrated the ability to detect diabetes risk in prediabetic individuals, helping doctors to personalise treatments.
"Ultimately, it's about giving people more awareness and control," says Quer. "Diabetes doesn't appear suddenly; it develops slowly, and now we have the tools to detect it earlier and intervene more intelligently."
[safe_summary] =>
)
)
)
[field_date_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_etichetta_box_lancio_news] => Array
(
)
[field_img_box_lancio_news] => Array
(
[und] => Array
(
[0] => Array
(
[fid] => 142067
[uid] => 26499
[filename] => ai_salute.jpg
[uri] => public://ai_salute.jpg
[filemime] => image/jpeg
[filesize] => 229888
[status] => 1
[timestamp] => 1756199852
[type] => image
[field_file_image_alt_text] => Array
(
)
[field_file_image_title_text] => Array
(
)
[field_folder] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2048
)
)
)
[metadata] => Array
(
[height] => 1322
[width] => 2519
)
[height] => 1322
[width] => 2519
[alt] => digital screen with images of the human brain
[title] =>
)
)
)
[field_link_alla_news] => Array
(
)
[field_link_esterno_news] => Array
(
)
[field_pagina_associata] => Array
(
)
[field_link_etichetta] => Array
(
)
[field_abstract_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
[format] =>
[safe_value] => Researchers, including alumni from the University of Padua, develop AI to detect diabetes risk through sensors, surpassing the limitations of the HbA1c test
)
)
)
[field_allegato_news] => Array
(
)
[field_categorie_news] => Array
(
[und] => Array
(
[0] => Array
(
[tid] => 2296
)
)
)
[field_pub_date] => Array
(
[und] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[value2] => 2026-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
)
[field_layout_news] => Array
(
[und] => Array
(
[0] => Array
(
[value] => single
)
)
)
[field_testo_opzionale_news] => Array
(
)
[field_url_en_page] => Array
(
[und] => Array
(
[0] => Array
(
[value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
[format] =>
[safe_value] => https://www.unipd.it/news/alumni-unipd-sviluppano-ia-rilevare-precocemente-diabete
)
)
)
[field_url_en_page_label] => Array
(
[und] => Array
(
[0] => Array
(
[value] => Italian version
[format] =>
[safe_value] => Italian version
)
)
)
[path] => Array
(
[pathauto] => 0
)
[name] => rossella.vezzosi
[picture] => 0
[data] => b:0;
[num_revisions] => 5
[current_revision_id] => 500796
[is_current] => 1
[is_pending] =>
[revision_moderation] =>
[entity_view_prepared] => 1
)
[#items] => Array
(
[0] => Array
(
[value] => 2025-08-26T00:00:00
[timezone] => Europe/Paris
[timezone_db] => Europe/Paris
[date_type] => date
)
)
[#formatter] => date_default
[0] => Array
(
[#markup] =>
Mar, 26/08/2025
)
)
)
