{"id":3243,"date":"2024-07-12T15:53:19","date_gmt":"2024-07-12T15:53:19","guid":{"rendered":"https:\/\/aifuturize.com\/?p=3243"},"modified":"2024-07-12T15:54:52","modified_gmt":"2024-07-12T15:54:52","slug":"harvards-ai-chatbot-outperforms-gpt-4v-with-90-tumor-detection","status":"publish","type":"post","link":"https:\/\/aifuturize.com\/vi\/ai-news\/harvards-ai-chatbot-outperforms-gpt-4v-with-90-tumor-detection\/","title":{"rendered":"Chatbot AI c\u1ee7a Harvard v\u01b0\u1ee3t tr\u1ed9i h\u01a1n GPT-4V v\u1edbi kh\u1ea3 n\u0103ng ph\u00e1t hi\u1ec7n kh\u1ed1i u 90%"},"content":{"rendered":"
\"\u0110\u01b0\u1eddng<\/span><\/div>

V\u01b0\u1ee3t qua GPT-4V! Chatbot AI c\u1ee7a Tr\u01b0\u1eddng Y Harvard n\u1ed5i b\u1eadt trong t\u1ef1 nhi\u00ean v\u1edbi \u0111\u1ed9 ch\u00ednh x\u00e1c ph\u00e1t hi\u1ec7n kh\u1ed1i u g\u1ea7n 90%<\/h2>\n

Hi\u1ec7n nay, ung th\u01b0 \u0111\u00e3 tr\u1edf th\u00e0nh m\u1ed9t trong nh\u1eefng nguy\u00ean nh\u00e2n g\u00e2y t\u1eed vong h\u00e0ng \u0111\u1ea7u tr\u00ean to\u00e0n th\u1ebf gi\u1edbi, v\u1edbi h\u00e0ng tri\u1ec7u ng\u01b0\u1eddi ch\u1ebft v\u00ec c\u0103n b\u1ec7nh n\u00e0y m\u1ed7i n\u0103m. T\u1ed5 ch\u1ee9c Y t\u1ebf Th\u1ebf gi\u1edbi \u0111\u00e3 tuy\u00ean b\u1ed1 r\u1eb1ng 1\/3 s\u1ed1 ca ung th\u01b0 c\u00f3 th\u1ec3 \u0111\u01b0\u1ee3c ch\u1eefa kh\u1ecfi n\u1ebfu ph\u00e1t hi\u1ec7n s\u1edbm v\u00e0 \u0111i\u1ec1u tr\u1ecb s\u1edbm.<\/p>\n

Tuy nhi\u00ean, t\u1ea7m so\u00e1t ung th\u01b0 lu\u00f4n l\u00e0 th\u00e1ch th\u1ee9c l\u1edbn trong l\u0129nh v\u1ef1c y t\u1ebf. \u0110\u1eb7c bi\u1ec7t, vi\u1ec7c x\u00e1c \u0111\u1ecbnh v\u00e0 ch\u1ea9n \u0111o\u00e1n ch\u00ednh x\u00e1c c\u00e1c kh\u1ed1i u l\u00e0 r\u1ea5t quan tr\u1ecdng \u0111\u1ec3 \u0111i\u1ec1u tr\u1ecb b\u1ec7nh nh\u00e2n trong ph\u00e2n t\u00edch b\u1ec7nh l\u00fd. Vi\u1ec7c kh\u00e1m b\u1ec7nh l\u00fd truy\u1ec1n th\u1ed1ng ph\u1ea7n l\u1edbn ph\u1ee5 thu\u1ed9c v\u00e0o chuy\u00ean m\u00f4n v\u00e0 ki\u1ebfn th\u1ee9c chuy\u00ean m\u00f4n c\u1ee7a b\u00e1c s\u0129 chuy\u00ean khoa.<\/p>\n

V\u1edbi s\u1ef1 ph\u00e1t tri\u1ec3n c\u1ee7a c\u00e1c m\u00f4 h\u00ecnh l\u1edbn nh\u01b0 GPT-4, vi\u1ec7c nghi\u00ean c\u1ee9u s\u1eed d\u1ee5ng tr\u00ed tu\u1ec7 nh\u00e2n t\u1ea1o (AI)<\/a> \u0111\u1ec3 h\u1ed7 tr\u1ee3 ch\u1ea9n \u0111o\u00e1n b\u1ec7nh l\u00fd \u0111\u00e3 d\u1ea7n xu\u1ea5t hi\u1ec7n. Tuy nhi\u00ean, nhi\u1ec1u h\u1ec7 th\u1ed1ng AI v\u1eabn g\u1eb7p ph\u1ea3i nh\u1eefng thi\u1ebfu s\u00f3t v\u1ec1 hi\u1ec7u su\u1ea5t v\u00e0 kh\u1ea3 n\u0103ng t\u01b0\u01a1ng t\u00e1c k\u00e9m trong c\u00e1c \u1ee9ng d\u1ee5ng th\u1ef1c t\u1ebf.<\/p>\n

G\u1ea7n \u0111\u00e2y, m\u1ed9t nh\u00f3m nghi\u00ean c\u1ee9u t\u1eeb Tr\u01b0\u1eddng Y Harvard v\u00e0 c\u00e1c c\u1ed9ng t\u00e1c vi\u00ean c\u1ee7a h\u1ecd \u0111\u00e3 ph\u00e1t tri\u1ec3n m\u1ed9t tr\u1ee3 l\u00fd AI t\u1ed5ng qu\u00e1t v\u1ec1 ng\u00f4n ng\u1eef h\u00ecnh \u1ea3nh cho b\u1ec7nh l\u00fd \u1edf ng\u01b0\u1eddi c\u00f3 t\u00ean PathChat. H\u1ec7 th\u1ed1ng n\u00e0y c\u00f3 th\u1ec3 x\u00e1c \u0111\u1ecbnh ch\u00ednh x\u00e1c b\u1ec7nh t\u1eeb c\u00e1c slide sinh thi\u1ebft trong g\u1ea7n 90% tr\u01b0\u1eddng h\u1ee3p, v\u01b0\u1ee3t tr\u1ed9i so v\u1edbi c\u00e1c m\u00f4 h\u00ecnh AI th\u00f4ng th\u01b0\u1eddng nh\u01b0 GPT-4V v\u00e0 c\u00e1c m\u00f4 h\u00ecnh y t\u1ebf chuy\u00ean nghi\u1ec7p hi\u1ec7n c\u00f3 tr\u00ean th\u1ecb tr\u01b0\u1eddng.<\/p>\n

B\u00e0i vi\u1ebft nghi\u00ean c\u1ee9u li\u00ean quan c\u00f3 ti\u00eau \u0111\u1ec1 \u201cM\u1ed9t phi c\u00f4ng ph\u1ee5 AI t\u1ea1o ra \u0111a ph\u01b0\u01a1ng th\u1ee9c cho b\u1ec7nh l\u00fd con ng\u01b0\u1eddi,\u201d<\/a> \u0111\u00e3 \u0111\u01b0\u1ee3c c\u00f4ng b\u1ed1 tr\u00ean t\u1ea1p ch\u00ed khoa h\u1ecdc Nature.<\/p>\n

\"M\u1ed9t

M\u1ed9t AI Copilot s\u00e1ng t\u1ea1o \u0111a ph\u01b0\u01a1ng th\u1ee9c cho b\u1ec7nh l\u00fd con ng\u01b0\u1eddi<\/p><\/div>\n

\n
\n
\n
\n

\u0110i\u1ec1u \u0111\u00e1ng ch\u00fa \u00fd l\u00e0 c\u00f4ng ngh\u1ec7 \u0111\u1ed9t ph\u00e1 n\u00e0y kh\u00f4ng ch\u1ec9 c\u00f3 th\u1ec3 x\u00e1c \u0111\u1ecbnh kh\u1ed1i u m\u00e0 c\u00f2n t\u01b0\u01a1ng t\u00e1c v\u1edbi ng\u01b0\u1eddi d\u00f9ng, cung c\u1ea5p c\u00e1c c\u00f4ng c\u1ee5 v\u00e0 quan \u0111i\u1ec3m m\u1edbi cho ch\u1ea9n \u0111o\u00e1n v\u00e0 nghi\u00ean c\u1ee9u b\u1ec7nh l\u00fd.<\/p>\n

PathChat: Tr\u1ee3 l\u00fd AI ph\u00e1t hi\u1ec7n b\u1ec7nh l\u00fd \u0111a ph\u01b0\u01a1ng th\u1ee9c<\/h3>\n

Trong nhi\u1ec1u n\u0103m, b\u1ec7nh l\u00fd t\u00ednh to\u00e1n \u0111\u00e3 \u0111\u1ea1t \u0111\u01b0\u1ee3c ti\u1ebfn b\u1ed9 \u0111\u00e1ng k\u1ec3 trong vi\u1ec7c ph\u00e2n t\u00edch d\u1eef li\u1ec7u h\u00ecnh th\u00e1i b\u1ec7nh l\u00fd v\u00e0 d\u1eef li\u1ec7u ph\u00e1t hi\u1ec7n ph\u00e2n t\u1eed. L\u0129nh v\u1ef1c nghi\u00ean c\u1ee9u th\u00edch h\u1ee3p n\u00e0y, \u0111\u01b0\u1ee3c h\u00ecnh th\u00e0nh nh\u1edd s\u1ef1 giao thoa gi\u1eefa b\u1ec7nh l\u00fd v\u1edbi c\u00f4ng ngh\u1ec7 AI v\u00e0 th\u1ecb gi\u00e1c m\u00e1y t\u00ednh, \u0111ang d\u1ea7n tr\u1edf th\u00e0nh \u0111i\u1ec3m n\u00f3ng nghi\u00ean c\u1ee9u trong ph\u00e2n t\u00edch h\u00ecnh \u1ea3nh y t\u1ebf.<\/p>\n

B\u1ec7nh l\u00fd t\u00ednh to\u00e1n li\u00ean quan \u0111\u1ebfn vi\u1ec7c s\u1eed d\u1ee5ng x\u1eed l\u00fd h\u00ecnh \u1ea3nh v\u00e0 c\u00f4ng ngh\u1ec7 AI \u0111\u1ec3 x\u00e2y d\u1ef1ng c\u00e1c m\u00f4 h\u00ecnh b\u1ec7nh l\u00fd t\u00ednh to\u00e1n AI. Nh\u1eefng m\u00f4 h\u00ecnh n\u00e0y thu \u0111\u01b0\u1ee3c h\u00ecnh \u1ea3nh m\u00f4 b\u1ec7nh h\u1ecdc v\u00e0 ti\u1ebfn h\u00e0nh \u0111\u00e1nh gi\u00e1 s\u01a1 b\u1ed9 v\u1ec1 h\u00ecnh th\u00e1i c\u1ee7a nh\u1eefng h\u00ecnh \u1ea3nh n\u00e0y \u0111\u1ec3 h\u1ed7 tr\u1ee3 ch\u1ea9n \u0111o\u00e1n, \u0111\u00e1nh gi\u00e1 \u0111\u1ecbnh l\u01b0\u1ee3ng v\u00e0 ra quy\u1ebft \u0111\u1ecbnh th\u00f4ng qua c\u00e1c k\u1ef9 thu\u1eadt ph\u00e2n t\u00edch h\u00ecnh \u1ea3nh t\u1ef1 \u0111\u1ed9ng.<\/p>\n

Hi\u1ec7n nay, v\u1edbi s\u1ef1 ph\u00e1t tri\u1ec3n b\u00f9ng n\u1ed5 c\u1ee7a c\u00e1c c\u00f4ng ngh\u1ec7 AI t\u1ed5ng qu\u00e1t \u0111\u01b0\u1ee3c \u0111\u1ea1i di\u1ec7n b\u1edfi ChatGPT, c\u00e1c m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef l\u1edbn \u0111a ph\u01b0\u01a1ng th\u1ee9c (MLLM) \u0111ang ng\u00e0y c\u00e0ng \u0111\u01b0\u1ee3c \u00e1p d\u1ee5ng trong nghi\u00ean c\u1ee9u b\u1ec7nh l\u00fd t\u00ednh to\u00e1n v\u00e0 th\u1ef1c h\u00e0nh l\u00e2m s\u00e0ng trong b\u1ec7nh l\u00fd. Tuy nhi\u00ean, trong l\u0129nh v\u1ef1c chuy\u00ean m\u00f4n cao v\u1ec1 b\u1ec7nh l\u00fd gi\u1ea3i ph\u1eabu, nghi\u00ean c\u1ee9u x\u00e2y d\u1ef1ng c\u00e1c tr\u1ee3 l\u00fd AI \u0111a ph\u01b0\u01a1ng th\u1ee9c, t\u1ed5ng qu\u00e1t cho b\u1ec7nh l\u00fd v\u1eabn \u0111ang \u1edf giai \u0111o\u1ea1n \u0111\u1ea7u.<\/p>\n

Trong c\u00f4ng vi\u1ec7c n\u00e0y, nh\u00f3m nghi\u00ean c\u1ee9u \u0111\u00e3 thi\u1ebft k\u1ebf m\u1ed9t tr\u1ee3 l\u00fd AI c\u00f3 kh\u1ea3 n\u0103ng t\u1ea1o ra \u0111a ph\u01b0\u01a1ng th\u1ee9c d\u00e0nh ri\u00eang cho nghi\u00ean c\u1ee9u b\u1ec7nh l\u00fd \u1edf ng\u01b0\u1eddi\u2014PathChat. H\u1ecd \u0111\u00e3 \u0111\u00e0o t\u1ea1o tr\u01b0\u1edbc h\u1ec7 th\u1ed1ng th\u00f4ng qua qu\u00e1 tr\u00ecnh h\u1ecdc t\u1ef1 gi\u00e1m s\u00e1t tr\u00ean h\u01a1n 100 tri\u1ec7u m\u1ea3nh h\u00ecnh \u1ea3nh m\u00f4 t\u1ebf b\u00e0o t\u1eeb h\u01a1n 1 tri\u1ec7u slide. B\u1eb1ng c\u00e1ch k\u1ebft h\u1ee3p \u0111i\u1ec1u n\u00e0y v\u1edbi b\u1ed9 m\u00e3 h\u00f3a h\u00ecnh \u1ea3nh thu\u1ea7n t\u00fay ti\u00ean ti\u1ebfn nh\u1ea5t, UNI, h\u1ecd \u0111\u00e3 t\u1ea1o ra m\u1ed9t MLLM c\u00f3 kh\u1ea3 n\u0103ng suy lu\u1eadn v\u1ec1 c\u1ea3 \u0111\u1ea7u v\u00e0o ng\u00f4n ng\u1eef h\u00ecnh \u1ea3nh v\u00e0 ng\u00f4n ng\u1eef t\u1ef1 nhi\u00ean. Sau khi tinh ch\u1ec9nh t\u1eadp d\u1eef li\u1ec7u g\u1ed3m h\u01a1n 450.000 \u0111i\u1ec3m d\u1eef li\u1ec7u h\u01b0\u1edbng d\u1eabn, PathChat \u0111\u00e3 \u0111\u01b0\u1ee3c x\u00e2y d\u1ef1ng.<\/p>\n

Nghi\u00ean c\u1ee9u cho th\u1ea5y PathChat kh\u00f4ng ch\u1ec9 c\u00f3 th\u1ec3 x\u1eed l\u00fd d\u1eef li\u1ec7u \u0111\u1ea7u v\u00e0o \u0111a ph\u01b0\u01a1ng th\u1ee9c m\u00e0 c\u00f2n ph\u1ea3n h\u1ed3i ch\u00ednh x\u00e1c c\u00e1c c\u00e2u h\u1ecfi li\u00ean quan \u0111\u1ebfn b\u1ec7nh l\u00fd ph\u1ee9c t\u1ea1p, x\u00e1c \u0111\u1ecbnh ch\u00ednh x\u00e1c b\u1ec7nh t\u1eeb c\u00e1c slide sinh thi\u1ebft trong g\u1ea7n 90% tr\u01b0\u1eddng h\u1ee3p.<\/p>\n

\"\u0110\u01b0\u1eddng

\u0110\u01b0\u1eddng d\u1eabn<\/p><\/div>\n

V\u01b0\u1ee3t qua GPT-4V v\u1edbi \u0111\u1ed9 ch\u00ednh x\u00e1c g\u1ea7n 90%<\/h3>\n

\u0110\u1ec3 ki\u1ec3m tra hi\u1ec7u su\u1ea5t ph\u00e1t hi\u1ec7n c\u1ee7a PathChat, nh\u00f3m nghi\u00ean c\u1ee9u \u0111\u00e3 so s\u00e1nh PathChat v\u1edbi m\u00f4 h\u00ecnh ngu\u1ed3n m\u1edf LLaVA, LLaVA-Med d\u00e0nh ri\u00eang cho y sinh v\u00e0 GPT-4V.<\/p>\n

H\u1ecd \u0111\u00e3 thi\u1ebft k\u1ebf th\u00ed nghi\u1ec7m so s\u00e1nh PathQABench, ph\u00e2n t\u00edch c\u00e1c tr\u01b0\u1eddng h\u1ee3p b\u1ec7nh l\u00fd t\u1eeb c\u00e1c c\u01a1 quan v\u00e0 ph\u01b0\u01a1ng ph\u00e1p th\u1ef1c h\u00e0nh kh\u00e1c nhau \u0111\u1ec3 so s\u00e1nh hi\u1ec7u su\u1ea5t ph\u00e1t hi\u1ec7n c\u1ee7a PathChat v\u1edbi LLaVA, LLaVA-Med v\u00e0 GPT-4V.<\/p>\n<\/div>\n

K\u1ebft qu\u1ea3 cho th\u1ea5y r\u1eb1ng d\u00f9 kh\u00f4ng cung c\u1ea5p b\u1ed1i c\u1ea3nh l\u00e2m s\u00e0ng, \u0111\u1ed9 ch\u00ednh x\u00e1c ch\u1ea9n \u0111o\u00e1n c\u1ee7a PathChat cao h\u01a1n \u0111\u00e1ng k\u1ec3 so v\u1edbi LLaVA 1.5 v\u00e0 LLaVA-Med. Khi ch\u1ec9 \u0111\u00e1nh gi\u00e1 h\u00ecnh \u1ea3nh, PathChat \u0111\u00e3 \u0111\u1ea1t \u0111\u01b0\u1ee3c \u0111\u1ed9 ch\u00ednh x\u00e1c 78,1% tr\u00ean t\u1ea5t c\u1ea3 c\u00e1c \u0111i\u1ec3m chu\u1ea9n t\u1ed5ng h\u1ee3p, cao h\u01a1n 52,4% so v\u1edbi LLaVA 1.5 v\u00e0 cao h\u01a1n 63,8% so v\u1edbi LLaVA-Med.<\/p>\n

V\u1edbi vi\u1ec7c \u0111\u01b0a v\u00e0o b\u1ed1i c\u1ea3nh l\u00e2m s\u00e0ng, \u0111\u1ed9 ch\u00ednh x\u00e1c c\u1ee7a PathChat t\u0103ng th\u00eam l\u00ean 89,5%, cao h\u01a1n 39,0% so v\u1edbi LLaVA 1.5 v\u00e0 cao h\u01a1n 60,9% so v\u1edbi LLaVA-Med.<\/p>\n

Th\u1eed nghi\u1ec7m so s\u00e1nh cho th\u1ea5y PathChat c\u00f3 th\u1ec3 thu \u0111\u01b0\u1ee3c kh\u1ea3 n\u0103ng d\u1ef1 \u0111o\u00e1n \u0111\u00e1ng k\u1ec3 ch\u1ec9 t\u1eeb c\u00e1c \u0111\u1eb7c \u0111i\u1ec3m h\u00ecnh \u1ea3nh c\u1ee7a h\u00ecnh \u1ea3nh, thay v\u00ec ch\u1ec9 d\u1ef1a v\u00e0o b\u1ed1i c\u1ea3nh l\u00e2m s\u00e0ng. N\u00f3 c\u00f3 th\u1ec3 s\u1eed d\u1ee5ng hi\u1ec7u qu\u1ea3 v\u00e0 linh ho\u1ea1t th\u00f4ng tin \u0111a ph\u01b0\u01a1ng th\u1ee9c \u0111\u1ec3 ch\u1ea9n \u0111o\u00e1n ch\u00ednh x\u00e1c h\u00ecnh \u1ea3nh m\u00f4 h\u1ecdc b\u1eb1ng c\u00e1ch k\u1ebft h\u1ee3p th\u00f4ng tin phi h\u00ecnh \u1ea3nh \u0111\u01b0\u1ee3c cung c\u1ea5p th\u00f4ng qua ng\u00f4n ng\u1eef t\u1ef1 nhi\u00ean th\u00f4ng th\u01b0\u1eddng.<\/p>\n

\u0110\u1ec3 \u0111\u00e1nh gi\u00e1 kh\u00e1ch quan t\u00ednh ch\u00ednh x\u00e1c c\u1ee7a t\u1eebng c\u00e2u tr\u1ea3 l\u1eddi c\u1ee7a m\u00f4 h\u00ecnh \u0111\u1ed1i v\u1edbi c\u00e1c c\u00e2u h\u1ecfi m\u1edf, nh\u00f3m nghi\u00ean c\u1ee9u \u0111\u00e3 tuy\u1ec3n d\u1ee5ng b\u1ea3y nh\u00e0 nghi\u00ean c\u1ee9u b\u1ec7nh h\u1ecdc \u0111\u1ec3 th\u00e0nh l\u1eadp m\u1ed9t h\u1ed9i \u0111\u1ed3ng \u0111\u00e1nh gi\u00e1. B\u1eb1ng c\u00e1ch so s\u00e1nh c\u00e2u tr\u1ea3 l\u1eddi c\u1ee7a b\u1ed1n m\u00f4 h\u00ecnh v\u1edbi 260 c\u00e2u h\u1ecfi m\u1edf, \u0111\u1ed9 ch\u00ednh x\u00e1c c\u1ee7a c\u00e1c ph\u00e1t hi\u1ec7n c\u1ee7a m\u00f4 h\u00ecnh \u0111\u00e3 \u0111\u01b0\u1ee3c ph\u00e2n t\u00edch.<\/p>\n

Cu\u1ed1i c\u00f9ng, \u0111\u1ed1i v\u1edbi c\u00e1c c\u00e2u h\u1ecfi m\u1edf m\u00e0 b\u1ea3y chuy\u00ean gia c\u00f3 th\u1ec3 \u0111\u1ea1t \u0111\u01b0\u1ee3c s\u1ef1 \u0111\u1ed3ng thu\u1eadn, \u0111\u1ed9 ch\u00ednh x\u00e1c t\u1ed5ng th\u1ec3 c\u1ee7a PathChat l\u00e0 78,7%, cao h\u01a1n l\u1ea7n l\u01b0\u1ee3t l\u00e0 26,4%, 48,9% v\u00e0 48,1% so v\u1edbi GPT-4V, LLaVA 1,5 v\u00e0 LLaVA-Med. Nh\u00ecn chung, PathChat th\u1ec3 hi\u1ec7n hi\u1ec7u su\u1ea5t v\u01b0\u1ee3t tr\u1ed9i so v\u1edbi ba m\u00f4 h\u00ecnh c\u00f2n l\u1ea1i.<\/p>\n

C\u00e1c nh\u00e0 nghi\u00ean c\u1ee9u ch\u1ec9 ra r\u1eb1ng PathChat c\u00f3 th\u1ec3 ph\u00e2n t\u00edch v\u00e0 m\u00f4 t\u1ea3 c\u00e1c chi ti\u1ebft h\u00ecnh th\u00e1i tinh vi trong h\u00ecnh \u1ea3nh m\u00f4 b\u1ec7nh l\u00fd. B\u00ean c\u1ea1nh \u0111\u1ea7u v\u00e0o h\u00ecnh \u1ea3nh, n\u00f3 c\u00f2n c\u00f3 th\u1ec3 tr\u1ea3 l\u1eddi c\u00e1c c\u00e2u h\u1ecfi y\u00eau c\u1ea7u v\u1ec1 b\u1ec7nh l\u00fd v\u00e0 ki\u1ebfn th\u1ee9c n\u1ec1n t\u1ea3ng y sinh n\u00f3i chung, khi\u1ebfn n\u00f3 tr\u1edf th\u00e0nh m\u1ed9t c\u00f4ng c\u1ee5 \u0111\u1ea7y h\u1ee9a h\u1eb9n \u0111\u1ec3 h\u1ed7 tr\u1ee3 c\u00e1c nh\u00e0 nghi\u00ean c\u1ee9u v\u00e0 b\u1ec7nh l\u00fd h\u1ecdc.<\/p>\n

M\u1eb7c d\u00f9 PathChat c\u00f3 hi\u1ec7u su\u1ea5t tuy\u1ec7t v\u1eddi trong c\u00e1c th\u1eed nghi\u1ec7m nh\u01b0ng n\u00f3 v\u1eabn ph\u1ea3i \u0111\u1ed1i m\u1eb7t v\u1edbi m\u1ed9t s\u1ed1 th\u00e1ch th\u1ee9c trong \u1ee9ng d\u1ee5ng th\u1ef1c t\u1ebf. Ch\u00fang bao g\u1ed3m vi\u1ec7c \u0111\u1ea3m b\u1ea3o m\u00f4 h\u00ecnh c\u00f3 th\u1ec3 x\u00e1c \u0111\u1ecbnh c\u00e1c truy v\u1ea5n kh\u00f4ng h\u1ee3p l\u1ec7 v\u00e0 tr\u00e1nh k\u1ebft qu\u1ea3 \u0111\u1ea7u ra sai s\u00f3t, duy tr\u00ec \u0111\u1ed3ng b\u1ed9 h\u00f3a v\u1edbi ki\u1ebfn th\u1ee9c y t\u1ebf m\u1edbi nh\u1ea5t v\u00e0 gi\u1ea3i quy\u1ebft th\u1ef1c t\u1ebf l\u00e0 d\u1eef li\u1ec7u \u0111\u00e0o t\u1ea1o c\u1ee7a PathChat ch\u1ee7 y\u1ebfu \u0111\u1ebfn t\u1eeb d\u1eef li\u1ec7u l\u1ecbch s\u1eed, c\u00f3 th\u1ec3 ph\u1ea3n \u00e1nh \u201cs\u1ef1 \u0111\u1ed3ng thu\u1eadn khoa h\u1ecdc trong qu\u00e1 kh\u1ee9\u201d thay v\u00ec th\u00f4ng tin m\u1edbi nh\u1ea5t.<\/p>\n

C\u00e1c nh\u00e0 nghi\u00ean c\u1ee9u tuy\u00ean b\u1ed1 r\u1eb1ng c\u00e1c nghi\u00ean c\u1ee9u trong t\u01b0\u01a1ng lai c\u00f3 th\u1ec3 n\u00e2ng cao h\u01a1n n\u1eefa kh\u1ea3 n\u0103ng c\u1ee7a PathChat, bao g\u1ed3m h\u1ed7 tr\u1ee3 to\u00e0n b\u1ed9 WSI gigapixel ho\u1eb7c nhi\u1ec1u \u0111\u1ea7u v\u00e0o WSI v\u00e0 t\u00edch h\u1ee3p nhi\u1ec1u h\u1ed7 tr\u1ee3 d\u00e0nh ri\u00eang cho nhi\u1ec7m v\u1ee5 h\u01a1n, ch\u1eb3ng h\u1ea1n nh\u01b0 \u0111\u1ebfm ch\u00ednh x\u00e1c ho\u1eb7c b\u1ea3n \u0111\u1ecba h\u00f3a \u0111\u1ed1i t\u01b0\u1ee3ng. Ngo\u00e0i ra, vi\u1ec7c t\u00edch h\u1ee3p PathChat v\u1edbi tr\u00ecnh xem slide k\u1ef9 thu\u1eadt s\u1ed1 ho\u1eb7c h\u1ed3 s\u01a1 y t\u1ebf \u0111i\u1ec7n t\u1eed c\u00f3 th\u1ec3 c\u1ea3i thi\u1ec7n h\u01a1n n\u1eefa t\u00ednh th\u1ef1c t\u1ebf c\u1ee7a n\u00f3 trong th\u1ef1c h\u00e0nh l\u00e2m s\u00e0ng.<\/p>\n

G\u1ea7n \u0111\u00e2y, m\u00f4 h\u00ecnh AI t\u1ea1o \u0111a ph\u01b0\u01a1ng th\u1ee9c PathChat 2 \u0111\u00e3 \u0111\u01b0\u1ee3c ph\u00e1t h\u00e0nh. N\u00f3 c\u00f3 th\u1ec3 suy lu\u1eadn v\u1ec1 h\u00ecnh \u1ea3nh v\u00e0 v\u0103n b\u1ea3n b\u1ec7nh l\u00fd, ch\u1ea5p nh\u1eadn \u0111\u1ea7u v\u00e0o xen k\u1ebd c\u1ee7a nhi\u1ec1u h\u00ecnh \u1ea3nh v\u00e0 v\u0103n b\u1ea3n c\u00f3 \u0111\u1ed9 ph\u00e2n gi\u1ea3i cao trong tr\u00ecnh xem slide t\u01b0\u01a1ng t\u00e1c, do \u0111\u00f3 cung c\u1ea5p c\u00e1c \u0111\u00e1nh gi\u00e1 to\u00e0n di\u1ec7n h\u01a1n cho t\u1eebng tr\u01b0\u1eddng h\u1ee3p t\u01b0 v\u1ea5n.<\/p>\n

So v\u1edbi PathChat 1, n\u00f3 \u0111\u00e3 c\u1ea3i thi\u1ec7n \u0111\u00e1ng k\u1ec3 hi\u1ec7u su\u1ea5t trong ch\u1ea9n \u0111o\u00e1n ph\u00e2n bi\u1ec7t v\u00e0 m\u00f4 t\u1ea3 h\u00ecnh th\u00e1i. N\u00f3 c\u0169ng c\u00f3 c\u00e1c kh\u1ea3 n\u0103ng n\u00e2ng cao trong c\u00e1c nhi\u1ec7m v\u1ee5 nh\u01b0 l\u00e0m theo h\u01b0\u1edbng d\u1eabn, tr\u1ea3 l\u1eddi c\u00e2u h\u1ecfi m\u1edf v\u00e0 t\u00f3m t\u1eaft b\u00e1o c\u00e1o.<\/p>\n